Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

Personality and Social Psychology

Review http://psr.sagepub.com/

A Review of the Tripartite Structure of Subjective Well-Being: Implications for Conceptualization,


Operationalization, Analysis, and Synthesis
Michael A. Busseri and Stan W. Sadava
Pers Soc Psychol Rev 2011 15: 290 originally published online 3 December 2010
DOI: 10.1177/1088868310391271

The online version of this article can be found at:


http://psr.sagepub.com/content/15/3/290

Published by:

http://www.sagepublications.com

On behalf of:

Society for Personality and Social Psychology

Additional services and information for Personality and Social Psychology Review can be found at:

Email Alerts: http://psr.sagepub.com/cgi/alerts

Subscriptions: http://psr.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://psr.sagepub.com/content/15/3/290.refs.html

>> Version of Record - Jun 28, 2011

OnlineFirst Version of Record - Dec 3, 2010

What is This?

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Personality and Social Psychology Review

A Review of the Tripartite Structure of 15(3) 290­–314


© 2011 by the Society for Personality
and Social Psychology, Inc.
Subjective Well-Being: Implications for Reprints and permission:
sagepub.com/journalsPermissions.nav

Conceptualization, Operationalization, DOI: 10.1177/1088868310391271


http://pspr.sagepub.com

Analysis, and Synthesis

Michael A. Busseri1 and Stan W. Sadava1

Abstract
Subjective well-being (SWB) comprises a global evaluation of life satisfaction and positive and negative affective reactions to
one’s life. Despite the apparent simplicity of this tripartite model, the structure of SWB remains in question. In the present
review, the authors identify five prominent structural conceptualizations in which SWB is cast variously as three separate
components, a hierarchical construct, a causal system, a composite, and as configurations of components. Supporting evidence
for each of these models is reviewed, strengths and weaknesses are evaluated, and commonalities and discrepancies among
approaches are described. The authors demonstrate how current ambiguities concerning the tripartite structure of SWB
have fundamental implications for conceptualization, measurement, analysis, and synthesis. Given these ambiguities, it is
premature to propose a definitive structure of SWB. Rather, the authors outline a research agenda comprising both short-
term and longer-term steps toward resolving these foundational, yet largely unaddressed, issues concerning SWB.

Keywords
well being, individual differences, structure

In 1984, an article by Ed Diener was published in Psycho- expanding body of research has provided a rich database
logical Bulletin that reviewed a (then) nascent body of addressing many important issues, including the reliability
research concerned with people’s subjective evaluations of and validity of self-report measures of SWB,1 stability and
their lives, which he referred to as “subjective well-being” malleability of SWB, correlates and predictors of SWB, and
(SWB). According to Diener, this growing area of inquiry national and societal-level differences in SWB (Eid &
had three hallmarks. First, the study of SWB was concerned Larsen, 2008). Studies indicate that higher (relative to lower)
with individuals’ subjective evaluations of their own lives levels of SWB are associated with fewer symptoms of men-
using whatever criteria they deemed appropriate. Second, tal illness, stronger interpersonal relations, more functional
SWB included people’s global assessment of their life over- health status, more adaptive dispositions and temperaments,
all. Third, the study of SWB encompassed both positive and more prosocial functioning, and more self-enhancing cogni-
negative appraisals rather than focusing exclusively on indi- tive styles (e.g., Diener, 1984, 1994, 2000; Diener, Suh,
cators of distress and dysfunction. Consistent with these Lucas, & Smith, 1999; Lyubomirsky, Sheldon, & Schkade,
hallmarks, Diener proposed that SWB composed two types 2005; Pressman & Cohen, 2005). Consequently, a high level
of components. The first component was a cognitive judg- of SWB has been conceptualized as an indicator of optimal
ment concerning one’s overall level of life satisfaction. The human functioning (e.g., Keyes, 2005; Ryan & Deci, 2001)
second type of component composed affective experiences, and is considered an important personal and societal goal by
reflecting people’s positive and negative emotional reactions both researchers and lay persons alike (Diener, 2000; Seligman,
to their lives. Thus, in Diener’s (1984) original formulation, 2000). Researchers have also advocated for the creation of
the three primary components of SWB were life satisfaction national accounts of well-being to facilitate monitoring SWB
(LS), positive affect (PA), and negative affect (NA).
This tripartite formulation of SWB has been widely
adopted. According to the PsycINFO database, the Diener 1
Brock University, St. Catharines, Ontario, Canada
(1984) article has been cited close to 1,000 times and thou-
Corresponding Author:
sands of empirical works have investigated SWB with Michael Busseri, Brock University, Department of Psychology, 500
respect to LS, PA, and NA, including more than 200 pub- Glenridge Avenue, St. Catharines, Ontario, Canada L2S 3A1
lished works on SWB from Diener himself. This rapidly Email: mbusseri@brocku.ca

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 291

within and across nations, the results of which could be used the structure of SWB, it is not possible at present to propose
to inform government policy and priorities (e.g., Carlisle & a single, optimal structure for SWB. Indeed, it is unlikely
Hanlon, 2007; Diener, Kesebir, & Lucas, 2008; Dolan & that any single study could achieve such a goal. Nonetheless,
White, 2006). Such signs are evidence that the study of SWB we believe that a nuanced review is timely and valuable for
is thriving, and with impact.2 several reasons. First, researchers engaged in the study of
Despite the large volume of SWB research based on SWB appear to be generally unaware that multiple diverging
Diener’s (1984) tripartite model, several fundamental issues structural conceptualizations exist and that uncertainty
remain in question, including the issue of primary focus in regarding the structure of SWB has major implications for
the present work—the structure of SWB. By structure, we the study of SWB. Even if some are aware of this issue, in
are referring to how SWB is internally organized with respect general SWB investigators have not attended to the unre-
to its three primary components, LS, PA, and NA. Simply solved ambiguities and discrepancies we describe. Rather, in
stated, our principal concern is how these three primary com- the typical empirical report investigators (including our-
ponents constitute, reflect, and/or combine to produce the selves in previously published works) adopt a particular
hypothetical construct called SWB. Insights concerning structural conceptualization of SWB without justification for
structure are critical to our understanding of SWB. At the or explicit consideration of the implications of this choice.
most basic level, knowledge concerning how the three SWB Consequently, research literatures based on each of the five
components interrelate informs our understanding of what structural models have proliferated largely independent of
SWB is. As we discuss in subsequent sections, various each other. The most important justification for the present
answers to this foundational issue have been proposed, work, therefore, is the pressing need to heighten awareness
including conceptualizing SWB as a global propensity to among SWB researchers, reviewers, and journal editors of
evaluate one’s life in a positive fashion, as a heterogeneous the fundamental importance of these issues to our under-
composite of ingredients, and even as a broad research area standing of SWB.
encompassing three distinct components. The structure of Second, it is our view that a review and evaluation of
SWB also has important implications for operationalization prominent structural models as well as the delineation of
and measurement (i.e., What should researchers be measur- unanswered questions and recommendations for directions
ing to assess SWB?), analysis (i.e., How should SWB be for future research will be of considerable heuristic value to
treated statistically with respect to its components?), and investigators new to the study of SWB. Third, investigation
conceptualization (i.e., What type of phenomenon is SWB?). of multidimensional phenomena and multifaceted constructs
Furthermore, clarity concerning the structure of SWB is crit- (e.g., attitudes, self-concept, personality) is not unique to the
ical to facilitating accurate interpretation and meaningful study of SWB, and the heterogeneity of structural approaches
synthesis of SWB-related research findings. to SWB is in many ways consistent with ambiguities noted
Although structure-related reviews of SWB have been in other domains of personality and social psychological
reported over the years concerning issues such as the relation research (e.g., Anderson & Sedikides, 1991; Carver, 1989;
between global LS and domain-specific satisfaction (e.g., Marsh & Yeung, 1998). Thus, although our primary goal is
Schimmack, 2008) and the link between PA and NA (e.g., to advance knowledge concerning the tripartite structure of
Schimmack & Crites, 2005), a systematic review of the SWB, we anticipate that the issues examined in the present
structure of SWB with respect to all three components has work will be informative to readers interested in a wide array
not been reported. In the present article, we review five alter- of topics.
native models of the structure of SWB as presented in the
empirical literature spanning the past 40 years of research
related to SWB. Empirical evidence pertaining to these Prominent Structural Models of SWB
structural models, strengths and weaknesses of each approach, Our literature review proceeded in an iterative fashion. We
and commonalities and discrepancies among the five con- began with the structure of SWB proposed by Diener (1984).
ceptualizations of SWB are delineated. Implications of the Using the PsycINFO database, we searched for all published
current state of ambiguity concerning the structure of SWB research on SWB authored or coauthored by Diener. Con-
are then described with respect to several issues of funda- ceptualizations concerning the structure of SWB and closely
mental importance to the study of SWB: measurement, anal- related concepts proposed by researchers prior to, and fol-
ysis and interpretation, and integration and synthesis of lowing, the publication of Diener’s (1984) seminal review
SWB-related findings. We then present a programmatic also were reviewed. To identify these relevant published
research agenda composing short-term and longer-term steps works, we searched the PsycINFO database for all published
that we anticipate will help resolve the ambiguity concerning research containing the following two combinations of key-
the structure of SWB. words: subjective well-being and structure or life satisfac-
Given the wide-ranging nature of these issues, coupled tion and positive affect and negative affect. This aspect of the
with a general lack of attention from researchers to clarifying literature search was supplemented by a review of the works

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


292 Personality and Social Psychology Review 15(3)

independently examined” (p. 127; also see Balatsky & Diener,


1993; Chwalisz, Diener, & Gallagher, 1988; Diener, 1994, 1996,
2008; Diener & Biswas-Diener, 2002; Diener & Emmons,
SWB 1985; Diener, Emmons, Larsen, & Griffin, 1985; Diener
et al., 1999; Diener, Gohm, Suh, & Oishi, 2000; Diener &
Lucas, 2000b; Diener, Lucas, & Napa Scollon, 2006; Diener,
Oishi, & Lucas, 2003; Diener & Suh, 1999; Diener, Suh,
LS Smith, & Shao, 1995; Lucas, Diener, & Suh, 1996; Oishi, Die-
ner, & Lucas, 2007; Pavot & Diener, 1993, 2004, 2008;
Pavot, Diener, Colvin, & Sandvik, 1991; Pavot, Diener, &
Fujita, 1990; Suh, Diener, Oishi, & Triandis, 1998).
PA
Diener’s (1984) tripartite division among SWB compo-
nents was consistent with early research on well-being in
which emphasis was given to the separability among what
NA were believed to be the primary components of well-being.
For example, based on results from large-scale social sur-
veys Campbell (1976, 1981) was among the first to pro-
pose that a sense of well-being encompassed LS, PA, and
NA, which were later reconceptualized as two major com-
Figure 1. Model 1—Subjective well-being (SWB) as three
ponents: satisfaction–dissatisfaction and happiness,
separate components defined as the balance between PA and NA.4 According to
Note: In this structural model, SWB refers to a domain of research Campbell, although these two dimensions were moderately
inquiry, comprising life satisfaction (LS), positive affect (PA), and negative interrelated (i.e., r = .50), each tapped different aspects of
affect (NA) as separate components. Correlations among the three SWB
life experience that do “not always move together” (Camp-
components, shown using dotted, double-headed arrows, are not of
primary interest in this model. bell, 1981, p. 38) and thus should be treated separately
(also see Andrews & Withey, 1976; Carp & Carp, 1982,
1983).
cited by Diener and colleagues as well as in the additional Consistent with the proposed separation among SWB
publications from other investigators. Several hundred arti- components, Diener and colleagues demonstrated that when
cles were identified. multiple measures of LS, PA, and NA were examined jointly
Below we describe five prominent structural models of in exploratory factor analysis, measures of each component
SWB identified from this collective body of published work. loaded strongly onto three separate components. For exam-
In Model 1, SWB is conceptualized as a broad research ple, Diener and Emmons (1985) reported that in a varimax-
domain composing three separate topics: LS, PA, and NA. In rotated factor analysis, multiple PA items (e.g., happy,
Model 2, SWB is cast as a hierarchical construct spanning pleased, joy) loaded onto a separate factor from multiple NA
three primary components and is modeled as a higher order items (e.g., angry, depressed), both of which loaded onto a
latent factor. In Model 3, SWB is approached as a composite separate factor from LS items (e.g., “the conditions of my life
experience reflected in the combination of LS, PA, and NA. are excellent”; also see Balatsky & Diener, 1993). Additional
In Model 4, SWB is conceptualized as a causal system in factor analytic evidence has accumulated over the past sev-
which PA and NA are treated as inputs to LS. Finally, in eral decades in support of the separability of LS, PA, and NA
Model 5, SWB is treated as an integrated system of compo- (e.g., Adler & Fagley, 2005; Heller, Komar, & Lee, 2007;
nents configured in distinct ways within individuals. Lucas, 2008; Lynn & Steel, 2006; McCullough, Huebner, &
Laughlin, 2000; Pavot, 2008; Sagiv & Schwartz, 2000; Sirgy
et al., 2006; Suh, 2002; Watkins, Woodward, Stone, & Kolts,
Model 1: SWB as Three 2003; Westerhof & Keyes, 2006). According to Westerhof,
Separate Components Thissen, Dittman-Kohli, and Stevens (2006), therefore, SWB
Description and Supporting Evidence. Diener (1984) described is a multidimensional construct in which LS, PA, and NA
LS, PA, and NA as different aspects that should be assessed “are related but clearly separable dimensions” (p. 100).
and examined separately to provide a full description of The proposed differentiation between the two affective
SWB (see Figure 1). This conceptualization has been components of SWB was also consistent with Bradburn’s
espoused by Diener in numerous publications over the past (1969; Bradburn & Caplovitz, 1965) pioneering research
25 years. Diener and Biswas-Diener (2002), for example, suggesting that PA and NA were orthogonal dimensions
suggested that “life satisfaction, pleasant affect, and lack rather than polar opposites. This conclusion was supported
of unpleasant affect are separable constructs that must be by subsequent early research from Diener and colleagues

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 293

(e.g., Diener & Emmons, 1985; Diener, Larsen, Levine, & content of the construct and a strategy for operationalization
Emmons, 1985; Diener, Sandvik, & Pavot, 1991) as well as (Borsboom, Mellenbergh, & van Heerden, 2003; Edwards &
more recent conceptual and empirical work concerning the Bagozzi, 2000). Operationalizing a psychological construct
structure of affective experience in which PA and NA are delin- typically requires specifying a set of observable (albeit
eated as orthogonal dimensions (e.g., Cacioppo, Gardner, & imperfect) indicators of the construct, the pattern of interre-
Berntson, 1999; Fredrickson, 2001; Tellegen, 1985; Watson, lations among which is used to infer the underlying latent
Weise, Vaidya, & Tellegen, 1999). Relevant evidence also construct (Bollen, 2002; Kline, 1998). Although the content
emerges from research in which the relation between PA and of the SWB construct is well defined in terms of its three
NA is found to differ in both degree and direction as a function primary components, the emphasis on the separateness of
of Western versus Eastern cultural identification (e.g., Peru- these components suggests that SWB is not a typical con-
novic, Heller, & Rafaeli, 2007; Schimmack, Diener, & Oishi, struct as its existence cannot be inferred from the relations
2002; Scollon, Diener, Oishi, & Biswas-Diener, 2005). Other among its three primary indicators. One possibility, there-
investigators have provided evidence suggesting that there may fore, is that SWB is simply a term that refers to a domain of
be fundamental differences between individuals with respect to research interest rather than a psychological construct.
their dispositional tendencies toward joint or opposing experi- Indeed, SWB has been described as a general domain of
ences of PA and NA (e.g., Bood, Archer, & Norlander, 2004; inquiry (Diener et al., 1999; Diener, Napa Scollon, & Lucas
Norlander, Bood, & Archer, 2002) and that positive and nega- 2003), as a broad category of phenomena (Pavot, 2008;
tive affective experiences may have dissociable implications for Pavot & Diener, 2004), and as a “generic term” with LS, PA,
cognition, motivation, and behavior (e.g., Ashby, Isen, & and NA as different variants of SWB (Howell & Howell, 2008,
Turken, 1999; Fredrickson, 2001; Storbeck & Clore, 2008). p. 537). Such explanations, however, do not explain why and
in what manner LS, PA, and NA are to be considered pri-
Commentary. From the perspective of the separate compo- mary components if SWB is not a psychological construct.
nents model, the broad goals of research on SWB are to A second and related issue concerns the meaning of the
delineate the causes and correlates of LS, PA, and NA as associations among LS, PA, and NA. As we review in subse-
separate phenomena, such that as our knowledge of these quent sections, ample empirical evidence shows that these
issues grows, so too does our understanding of SWB. Conse- associations are often substantial and robust. It is unclear
quently, each of the three primary dimensions of SWB can from the perspective of the separate components model,
be examined separately (Pavot, 2008). Consonant with this however, what is to be made of these correlations among LS,
logic, a large volume of empirical information has accrued PA, and NA. One possibility is that the shared variance
concerning the correlates and predictors of LS, PA, and NA among SWB components is simply a methodological artifact
(DeNeve & Cooper, 1998; Diener, 1984, Diener et al., 1999; resulting from the use of self-report ratings. Alternatively,
Steel, Schmidt, & Shultz, 2008). This research includes the covariance among LS, PA, and NA may index a global
some studies suggesting that LS, PA, and NA may have dif- cognitive bias or predisposition toward viewing one’s life is
ferent nomological networks. For example, in studies exam- a positive manner (Cummins & Nistico, 2002; Robinson et al.,
ining all three SWB components jointly, extraversion and 2003). Proponents of the separate components model (e.g.,
positive social interactions are uniquely positively related to Andrews & Robinson, 1991; Argyle & Martin, 1991; Eid,
PA and LS (e.g., Costa & McCrae, 1980; DeNeve & Cooper, 1997; Elliot, Gable, & Mapes, 2006; Larsen & Eid, 2008)
1999; Heller, Watson, & Ilies, 2006), whereas neuroticism have not addressed the implications of the correlations
and depression typically show more substantial relations among components with respect to the meaning and struc-
with NA (DeNeve & Cooper, 1999; Watson, Clark, & Tel- ture of SWB. It is unclear, therefore, how to interpret asso-
legen, 1988). Related, Schimmack, Schupp, and Wagner ciations among LS, PA, and NA from the perspective of the
(2008) showed that socioeconomic conditions (e.g., unem- separate components model. Consequently, it is uncertain
ployment) had a stronger predictive effect on LS than on PA whether an exclusive emphasis on the distinctiveness of
and NA, whereas personality traits (e.g., extraversion, neu- these components can be reconciled with other prominent
roticism) had a stronger effect on PA and NA than on LS. A structural models (presented below) in which associations
recent addition to this literature is studies examining all three among the three SWB components are a defining feature.
SWB components as separate but simultaneous predictors of
other outcomes. Zelenski, Murphy, and Jenkins (2008), for
example, found that worker productivity was uniquely pre- Model 2: SWB as Hierarchical Construct
dicted by PA, rather than NA or LS. Description and Supporting Evidence. Despite the regularity
There are important challenges attending the separate with which the separability of LS, PA, and NA components
components model of SWB. One issue pertains to the status has been emphasized, the commonality among SWB com-
of SWB as a psychological phenomenon. Defining a psycho- ponents also has been highlighted. In early research on
logical construct requires specification of both the conceptual well-being, some investigators reported strong positive

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


294 Personality and Social Psychology Review 15(3)

correlations between global ratings of LS and positive (vs.


negative) affective experience (e.g., Campbell, Converse, &
Rodgers, 1976; Kammann, Farry, & Herbison, 1984), sug-
gesting the presence of a “common underlying construct”
(Lohmann, 1977, p. 74). In support of this notion, Larsen, LS +
Diener, and Emmons (1985) reported factor analytic evi-
dence indicating that measures of LS, happiness,3 and affect
+
loaded moderately to strongly onto a single factor (ls rang- PA SWB
ing in magnitude from .40 to .87). Other investigators also
have employed exploratory factor analysis to show that mea-

sures of LS, PA, and NA all load strongly onto a single latent
NA
component (e.g., McNeil, Stones, & Kozma, 1986; Sheldon
et al., 2004; Sheldon & Hoon, 2007; Sheldon & Lyubomirsky,
2006). Collectively, these studies suggest that the satisfaction
and affect components of SWB are “two sides of the same
coin” (Biswas-Diener, Vitterso, & Diener, 2005, p. 221) that
Figure 2. Model 2—Subjective well-being (SWB) as a hierarchical
form “a strong, general, subjective well-being factor” (Suh et construct
al., 1998, p. 484; also see Diener, 1994, 1995; Diener, Diener, Note: In this structural model, SWB is conceptualized as a latent higher
& Diener, 1995; Diener et al., 1999; Diener, Sandvik, Pavot, & order factor, reflected in three first-order indicators: life satisfaction
Gallagher, 1991; Diener, Suh, & Oishi, 1997; Kim-Prieto, (LS), positive affect (PA), and negative affect (NA). Associations among
the three first-order components are assumed to be a result of (and
Diener, Tamir, Scollon, & Diener, 2005; Pavot & Diener, 1993, thus explained by) latent SWB. Factor loadings of the three indicators on
2008; Seidlitz, Wyer, & Diener, 1997; Tov & Diener, 2007). the latent SWB construct, shown by directional arrows, are assumed to
Consistent with this view, a composite SWB score is some- be substantive. + indicates an anticipated positive loading; – indicates an
anticipated negative loading.
times derived as the combination of LS, PA, and reverse-
scored NA (e.g., Elliot et al., 2006; Malka & Chatman, 2003;
Nes, Roysamb, Tambs, Harris, & Reichborn-Kjennerud, Molnar, Busseri, Perrier, & Sadava, 2009; Vitterso, 2001).
2006; Sheldon et al., 2004; Sheldon & Hoon, 2007; Sheldon & Based on such evidence, many other researchers also have
Lyubomirsky, 2006; Vitterso, 2001, 2004). endorsed a hierarchical conceptualization of SWB (e.g., Busseri,
Expanding this conceptualization of a unitary SWB, Sadava, & DeCourville, 2007; Chamberlain, 1988; Ho &
researchers have advocated for a hierarchical structure com- Cheng, 2007; Kozma, Stone, & Stones, 2000; Malka &
prising a higher order SWB factor and three lower order Chatman, 2003; McCulloch, 1991; McNeil et al., 1986; Nes
components (see Figure 2). For example, in early research on et al., 2006; Robinson, Crawford Solberg, Vargas, & Tamir,
well-being, Liang (1984, 1985) provided confirmatory factor 2003; Sheldon & Hoon, 2007; Sheldon & Lyubomirsky,
analytic evidence of a higher order latent well-being factor 2006; Shmotkin & Hadari, 1996; Vitterso, 2004; Vitterso &
with strong loadings from latent first-order factors, including Nilsen, 2002; Wallace & Bergeman, 2007; Wiese, 2007).
LS and affect (also see L. K. George, 1981; Lawton, 1982, Importantly, although substantial loadings of the first-
1983a, 1983b; Liang & Bollen, 1983, 1985; Stones & Kozma, order components on the higher order SWB factor are typi-
1980, 1985). The hallmark of these early hierarchical models cally observed, the magnitudes of these factor loadings often
was a superordinate well-being construct thought to be respon- fall well short of unity, suggesting substantial variance spe-
sible for the correlations among various first-order compo- cific to LS, PA, and NA (i.e., independent of the higher order
nents that typically included LS, PA, and NA. Similarly, SWB factor), as demonstrated in the example provided above
Diener and colleagues have described SWB in terms of both from Arthaud-Day et al. (2005; also see Busseri et al., 2007;
global well-being and distinguishable LS, PA, and NA com- Vitterso, 2004). Consequently, rather than focusing exclu-
ponents (e.g., Diener, Napa Scollon, & Lucas, 2003; Diener & sively on the latent higher order SWB factor, some research-
Tov, 2007; Larsen et al., 1985; Oishi et al., 2007). ers employing a hierarchical construct approach have emphasized
In support of these notions, confirmatory factor analysis the importance of examining both the higher order SWB fac-
has been employed to show that a latent second-order SWB tor and the lower order LS, PA, and NA components (e.g.,
factor can explain the covariance among the three latent first- Arthaud-Day et al., 2005; Busseri et al., 2007; Chamberlain,
order factors. For example, Arthaud-Day, Rode, Mooney, 1988; McCulloch, 1991).
and Near (2005) demonstrated that in multiple samples, a
higher order latent SWB factor had factor loadings from first- Commentary. From the perspective of a contemporary hierar-
order LS, PA, and NA factors ranging in absolute magnitude chical construct structural conceptualization, a complete
from .18 to .77 (Mdn = .32; also see Lawrence & Liang, account of SWB requires attention to both a global SWB fac-
1988; Linley, Maltby, Wood, Osborne, & Hurling, 2009; tor and three first-order factors. In several studies, researchers

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 295

have examined correlates or predictors of a composite SWB Even if consistent evidence concerning the generaliz-
score or a latent higher order SWB factor along with separate ability of a latent higher order SWB factor were found,
LS, PA, and NA scores (e.g., Elliot et al., 2006; Sheldon et al., such evidence would not clarify the meaning of the latent
2004; Sheldon & Lyubomirsky, 2006; Vitterso, 2001, 2004; factor. A higher order factor spanning LS, PA, and NA
Wallace & Bergeman, 2007). For example, Vitterso (2001) components has been interpreted as an underlying, global
found that extraversion (controlling for neuroticism) was sense of satisfaction and happiness with one’s life (e.g.,
uniquely related to PA but not with LS, NA, or a composite Andrews & Withey, 1976; Arthaud-Day et al., 2005; Liang,
SWB score. If such a hybrid structural approach was investi- 1985; Seidlitz et al., 1997), a personal tendency toward
gated systematically across studies, the nomological net- positive (vs. negative) appraisals for one’s life along a
works of the higher order SWB factor and each of the three global good–bad dimension (Cummins & Nistico, 2002;
first-order components could be evaluated simultaneously Diener, Napa Scollon, Oishi, Dzokoto, & Suh, 2000; Oishi
(Busseri et al., 2007). Ultimately, the tabulation of such find- & Diener, 2001; Vitterso, 2004; Vitterso & Nilsen, 2002)
ings across studies could provide extensive information con- similar to a positivity halo (Schimmack et al., 2008), and a
cerning the commonalities and dissociations between a reflection of one’s beliefs about oneself and one’s emotions
higher order SWB factor and its first-order LS, PA, and NA rather than an accurate portrayal one’s life to date (Robin-
components. son et al., 2003; Robinson et al., 2004). The variety of these
An important unresolved issue concerning the hierarchi- proposals highlights an ambiguity concerning the meaning
cal construct conceptualization pertains to the nature of the of the common variance among SWB components—argu-
associations among SWB components. Without positive ably the central feature of the hierarchical construct struc-
relations between LS and PA and negative relations between tural conceptualization.
NA and both LS and PA, a latent factor model of SWB is not
tenable. Although reviewers have commented on SWB-
related findings concerning associations between PA and Model 3: SWB as a Causal System
NA (Schimmack, 2008), an empirical review encompassing Description and Supporting Evidence. Another prominent struc-
relations among all three SWB components has yet to be tural approach conceptualizes SWB as a system of causally
reported. In some studies, including the factor analytic stud- related components. Specifically, LS is treated as an out-
ies summarized above, substantive positive correlations come of PA and NA, wherein PA and NA make independent
between LS and PA and substantive negative correlations contributions to LS (see Figure 3). For example, Bradburn
between LS and NA have been observed. However, the asso- (1969) conceptualized well-being as resulting from the rela-
ciation between PA and NA varies across studies and appears tive strengths of orthogonal PA and NA factors, such that the
to be moderated by several factors including the rating scale joint experience of high PA and low NA was anticipated to
employed, affect descriptors used, age of the sample, and result in the most positive life evaluations. Similarly, Costa
cultural group surveyed (Schimmack & Crites, 2005). Con- and McCrae (1980) proposed that positive and negative
sequently, the viability and generalizability of a hierarchical affective experiences are “subjectively balanced by an indi-
conceptualization would appear to hinge on demographic vidual to arrive at a net sense of subjective well-being” (p. 675;
and methodological factors, including the type of measure- also see Brenner, 1975; Kozma & Stones, 1980). Further-
ment instrument employed and the culture of the respon- more, according to Diener and colleagues, people use infor-
dents. More generally, a long-standing debate persists mation about positive and negative affective experiences
concerning the structure of positive and affective experience when constructing global SWB judgments such that long-
as orthogonal factors (e.g., Cacioppo et al., 1999; Fredrickson, term affect “accounts for half of the variability” in LS
2001; Watson et al., 1999) versus opposing ends of a single (Kim-Prieto et al., 2005, p. 284; also see Diener, 1994;
affect valence continuum (e.g., Russell & Barrett, 1999). At Diener, Napa Scollon, & Lucas, 2003; Diener, Lucas, Oishi,
present, therefore, the tenability of a hierarchical SWB con- & Suh, 2002; Emmons & Diener, 1985). For example,
struct based on a consistent pattern of correlations among Schimmack et al. (2002) reported that across samples of var-
LS, PA, and NA is uncertain. ious cultural groups, PA and NA each contributed to a “hedonic
An additional issue is the causal status of a higher order balance” factor, which then predicted LS (standardized path
latent SWB factor. A hierarchical construct model assumes coefficients ranged from .48 to .76).
that the higher order latent factor is responsible for produc- Other researchers also employ a causal system approach
ing the common variance among LS, PA, and NA and should in which PA and NA are treated as predictors of LS (e.g.,
determine (at least in part) the variability in all three compo- Cheng, 2004; Davern & Cummins, 2006; Davern, Cummins,
nents. To date, however, these implications have not been & Stokes, 2007; J. M. George, 1991; Klonowicz, 2001;
systematically tested despite their fundamental importance Kozma, Stone, Stones, Hannah, & McNeil, 1990; Larsen &
to establishing the validity of a hierarchical conceptualiza- Prizmic, 2008; Lent, 2004; McKennell, 1978; Schimmack,
tion of the tripartite structure of SWB. 2008; Schimmack et al., 2002, 2008; Singh & Jha, 2008). In

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


296 Personality and Social Psychology Review 15(3)

this parsimony, the causal system model does not require or


constrain the PA–NA correlation to a particular direction or
magnitude and, therefore, is unhindered by the ongoing
debate concerning PA and NA as opposing versus orthogo-
SWB nal dimensions of affective experience. Rather, the relative
contributions of PA and NA to predicting LS can be esti-
mated simultaneously, regardless of the nature of the relation
between the two predictors.
PA
+ Despite these apparent advantages of conceptualizing the
tripartite structure of SWB in terms of a causal system,
important implications remain unaddressed. Using the term
LS SWB to refer to a system of relations among three compo-
– nents creates several ambiguities. For example, it is unclear
whether all three components need to be included in a given
study or whether research examining LS without PA and NA
NA would adequately inform SWB. It is not uncommon for
investigators to frame their studies as explorations of SWB
when only the LS component is measured (e.g., Lucas, 2007;
Napier & Jost, 2008), suggesting that some researchers view
Figure 3. Model 3—Subjective well-being (SWB) as a causal SWB as interchangeable or synonymous with LS judgments.
system Perhaps, therefore, the causal system conceptualization is
In this structural model, SWB comprises a series of causal relations
among components. Specifically, positive affect (PA) and negative affect not actually a three-component model of SWB in which LS,
(NA) are conceptualized as inputs to life satisfaction (LS), as indicated PA, and NA are primary indicators of SWB per se but rather
by the pair of directional arrows. + indicates an anticipated positive a single-component model comprising LS in which PA and
path coefficient; – indicates an anticipated negative path coefficient. The
correlation between PA and NA, shown by a dotted, double-headed
NA are secondary (predictive) factors. Thus, it is unclear
arrow, is not a key feature of the model. what SWB refers to in the causal system model and what
status PA and NA have with respect to SWB in this model
(e.g., as indicators vs. predictors).
some versions of this model, satisfaction with one’s life is Another limitation is that researchers affirming the causal
considered to be “the essence” of SWB (Davern al., 2007, p. system model have not provided evidence that PA and NA
432; also see Oishi & Koo, 2008). In other models, SWB is cause LS. Instead, studies examining PA and NA as predic-
used in reference to the system of variables rather than just tors of LS have employed cross-sectional correlational
the LS component (e.g., Schimmack et al., 2002). A com- designs in which all three SWB components are measured
mon rationale for a causal system approach is the proposition concurrently. Direct evidence has yet to be provided show-
that people rely on affective information (e.g., how they feel ing that PA and NA cause LS or even predict changes in LS
about their lives and/or the balance of positive versus nega- over time or across situations. Consequently, the defining
tive emotions reactions or memories) when forming assess- feature of the causal system structural model of SWB
ments of overall LS (e.g., Kozma et al., 1990; Kuppens, remains unsubstantiated.
Realo, & Diener, 2008; Schimmack, 2008; Schimmack
et al., 2002, 2008). Other researchers have cast PA and NA as
indicators of basic affective dispositions that influence the Model 4: SWB as a Composite
types of personal experiences and reactions to life events Description and Supporting Evidence. A fourth structural model
and, ultimately, affect LS judgments (J. M. George, 1991; specifies all three components as inputs to SWB. Beiser
Kozma et al., 1990). Thus, across causal system structural (1974), for example, proposed that one’s general sense of
conceptualizations, PA and NA are thought to directly influ- well-being is determined by positive and negative affective
ence LS, either because of the impact of affective informa- reactions as well as a cognitive judgment concerning overall
tion on satisfaction judgments or as a result of dispositional contentment with one’s life. In this study, measures of LS,
tendencies toward positive and negative affective experi- PA, and NA each contributed uniquely to the prediction of a
ences and reactions to one’s life. global sense of well-being (prs = .30, .31, and –.30, respec-
tively, ps < .001). Andrews and Crandall (1976) advanced a
Commentary. A potential advantage of this structural model is similar model and delineated three separate factors to explain
the delineation of just one key criterion: a highly satisfying variance in evaluations about one’s life as a whole: PA, NA,
life. Accordingly, much could be learned about SWB from and cognition (Andrews & McKennell, 1980; Andrews &
studying the correlates and predictors of LS. In addition to Withey, 1976; McKennell & Andrews, 1980). Thus, in these

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 297

early models, global life assessments were thought to result


from affective experiences and cognitive evaluations.
Structural models of this sort are consistent with a com-
posite view of SWB. A composite approach assumes that the
combination of LS, PA, and NA is needed to fully capture LS +
the notion of SWB because LS, PA, and NA are joint con-
tributors to one’s overall sense of well-being. From this per-
spective, all three components (LS, PA, and NA) are required +
to assess SWB as the absence of any one component would
PA SWB
necessarily omit a key ingredient. Thus, a composite –
approach conceptualizes the structure of SWB in terms of a
simple recipe composing a blend of LS, PA, and NA. NA

Commentary. Composite scores are appropriate in several situ-


ations, including those in which the assumptions of the stan-
dard latent factor approach do not hold (Bollen & Lennox,
1991). Typically, a latent factor is modeled as if it were the Figure 4. Model 4—Subjective well-being (SWB) as a composite
cause of its indicators, that is, the indicators are assumed to Note: In this structural model, SWB is conceptualized as a recipe, formed
through the combination of three causal indicators (i.e., ingredients), life
reflect the latent factor and the covariance among indicators is
satisfaction (LS), positive affect (PA), and negative affect (NA), as reflected
used to infer the underlying latent factor (Kline, 1998). In the by the three directional arrows. + indicates an anticipated positive
study of personality traits, for example, extraversion is typi- path coefficient; – indicates an anticipated negative path coefficient.
cally conceptualized as a latent (i.e., unmeasured) tendency Correlations among the three causal indicators, shown by the dotted,
double-headed arrows, are not of primary interest in this model.
inferred from the co-occurrence of observable behaviors and
characteristics such as talkativeness and excitement seeking
(McCrae & John, 1992). However, this standard approach indicators (Bollen & Lennox, 1991; MacCallum & Browne,
assumes that the indicators are interchangeable manifesta- 1993; MacKenzie, Podsakoff, & Jarvis, 2005). Accordingly, a
tions of the latent construct, that a change in the latent con- latent composite SWB variable could be modeled as resulting
struct would change each indicator in a similar fashion, and from the combination of its three primary ingredients, that is,
that the indicators share the same set of antecedents and con- LS, PA, and NA (see Figure 4). If the latent composite SWB
sequences (Jarvis, MacKenzie, & Podsakoff, 2003). As we variable were estimated using constrained path coefficients of 1,
have seen, there is no consensus among SWB researchers 1, and -1 from standardized scores for LS, PA, and NA, respec-
concerning the appropriateness of these assumptions. tively, the resulting latent composite variable would be mathe-
In contrast, a “latent composite variable” (Bollen & Lennox, matically equivalent to a measured composite score formed by
1991; MacCallum & Browne, 1993) is thought to be pro- averaging or summing standardized LS, PA, and (reverse-
duced by a set of “causal indicators” (Bollen & Lennox, scored) NA scores.
1991). That is, the indicators are thought to produce the As we have noted, however, investigators assuming a
latent variable, rather than vice versa. A latent composite hierarchical construct structural model (Model 2) have
approach assumes that the indicators are not equivalent man- employed composite SWB scores (typically a unit-weighted
ifestations of the same latent construct, that a change in the combination of standardized scores for LS, PA, and reverse-
composite would not necessarily change the indicators, that scored NA) as stand-ins for a higher order latent SWB factor
the indicators do not necessarily share the same set of ante- (e.g., McNeil et al., 1986; Sheldon & Hoon, 2007). Method-
cedents and consequences, and that change in any one causal ologists have emphasized that composites of this sort do not
indicator would create a change in the composite (Edwards provide consistent or valid estimates of an underlying latent
& Bagozzi, 2000; MacCallum & Browne, 1993). For exam- factor assumed by the hierarchical construct approach (e.g.,
ple, higher (vs. lower) socioeconomic status (SES) can be Bentler & Kano, 1990; Bollen & Lennox, 1991; Grice, 2001;
thought of as emerging from a combination of education Steiger, 1990). Therefore, although studies examining cor-
level, occupation type, and income level such that a change relates and predictors of measured composite SWB scores
in any one of these three causal indicators produces a change are typically based on the assumption of a hierarchical SWB
in SES level, rather than vice versa (Bollen & Lennox, 1991). construct (i.e., Model 2), such investigations actually pro-
Importantly, if paths from each causal indicator to the latent vide information concerning a composite structural model
composite variable are given equal weights (assuming a com- (i.e., Model 4) in which SWB is produced by (rather than
mon scaling metric across indicators), then the resulting latent manifested in) its three main causal ingredients.
composite variable is identical to a measured composite score Another consideration is that through the use of structural
formed by averaging or summing the same set of unit-weighted equation modeling (SEM), both the latent composite SWB

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


298 Personality and Social Psychology Review 15(3)

variable and each of the three causal indicators can be exam-


ined in relation to other variables of interest. Because the
composite can be estimated as a latent variable created by its High SWB Low LS
indicators rather than as a measured computed score (which
would be statistically dependent on the set of its indicators if high LS low LS
analyzed in the same analytic model), both the latent com-
posite variable and the set of causal indicators can be exam-
high PA moderate PA
ined in the same model. Consequently, predictors of a latent
composite SWB variable could be assessed and unique rela-
tions with each of the causal indicators (i.e., LS, PA, and low NA moderate NA
NA) could be tested simultaneously. Through employing
such a method, information concerning the nomological net-
work of SWB as a composite could be explored alongside
Figure 5. Model 5—Subjective well-being (SWB) as a
LS, PA, and NA as separate ingredients.
configuration of components
Furthermore, although a latent composite variable model Note: In this structural model, SWB is conceptualized as an integrated
assumes that LS, PA, and NA each contribute to the SWB system of components, organized within individuals in terms of
composite in a particular statistical direction (i.e., positivity, distinct configurations of life satisfaction (LS), positive affect (PA), and
negative affect (NA). No particular pattern of correlations among SWB
positivity, and negativity, respectively), this model makes no
components is assumed. Rather, individuals, or subgroups of individuals,
claims concerning the nature of the correlations among the are expected to differ with respect to their SWB configurations, which
three causal indicators themselves. Indeed, the common could include congruous and incongruous profiles, examples of which are
variance among components is not important in this particu- illustrated, respectively, by the “high SWB” and “low LS” configurations.
lar conceptualization; rather, emphasis is given to the combi-
nation of components irrespective of the degree to which
these components are themselves interrelated. Consequently, Model 5: SWB as a Configuration
a composite conceptualization of the structure of SWB can of Components
accommodate various patterns of correlation among LS, PA,
and NA—rather than assuming independence (as in Model 1) Description and Supporting Evidence. A fifth structural concep-
or requiring substantial correlations among at least some tualization focuses on how LS, PA, and NA are configured
components (as in Model 2 and Model 3). within individuals. Since the beginning of research on SWB,
The flip side to this flexibility, however, is that it is unclear “high SWB” has been described as the within-individual
how to interpret the observed correlations among LS, PA, and combination of high LS, frequent PA, and infrequent NA
NA components with respect to a latent composite SWB. As (e.g., Andrews & Withey, 1976; Diener, 1984). Various
with the separate components model (Model 1), although this other combinations of SWB components also have been pro-
common variance may simply reflect naturally occurring posed. Diener and Lucas (1999), for example, speculated
covariation, it could also be interpreted as a methodological that although persons reporting high levels of pleasant affect
artifact (e.g., resulting from exclusive reliance on self-report) and little unpleasant affect could be described as “happy,”
or a positive self-evaluation halo reflecting a general tendency someone experiencing high levels of both pleasant and
toward evaluating one’s life positively. Both interpretations, unpleasant affect might be labeled “highly emotional.” Oth-
however, assume that the meaning of the covariation among ers might be satisfied with their lives despite experiencing
LS, PA, and NA in a composite model of SWB (i.e., Model 4) infrequent PA and frequent NA or dissatisfied with their
is not informative with respect to SWB. lives despite frequent PA and infrequent NA (Arthaud-Day
Furthermore, a potentially confusing implication of a et al., 2005). Even individuals reporting similar overall lev-
composite model is that a latent composite SWB variable in els of SWB may be characterized by distinct hedonic profiles
which the paths from standardized scores for LS, PA, and (Lyubomirsky et al., 2005). These proposals all emphasize
NA were fixed to 1, 1, and –1, respectively, would be statis- combinations of SWB components as they occur within
tically equivalent to a unit-weighted measured SWB com- individuals.
posite score. As we have seen, this latter approach is Consistent with these notions, Shmotkin (2005) defined
typically employed by researchers conceptualizing SWB as SWB as an integrated system of components in which LS,
a hierarchical construct (Model 2), despite cautions from PA, and NA are organized within individuals in terms of dis-
methodologists. Thus, utilization of a latent composite tinct configurations (see Figure 5). Rather than fixed disposi-
SWB approach by proponents of a composite structural tions, SWB configurations are conceptualized in Shmotkin’s
conceptualization (Model 4) may create confusion regard- model as flexible and adaptive modes of experience that
ing the meaning and interpretation of results based on a allow individuals to maximize positive functioning in the face
composite SWB score. of adversity or threat. In support of a configuration model, a

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 299

small number of studies has differentiated distinct subgroups nature of the relations among LS, PA, and NA. Rather, a
of individuals using cross-tabulations of pairs of SWB rat- configural perspective is consistent with a growing body of
ings, typically global assessments of LS and happiness (e.g., evidence suggesting that the relation between PA and NA,
McKennell, 1978; Michalos, 1980; Shmotkin, 1998; Shmotkin, and thus the structure of SWB, may vary both between indi-
Berkovich, & Cohen, 2006). Other researchers have used viduals and within individuals over time. According to the
cut scores to define high versus low SWB groups based on evaluative space model (Cacioppo et al., 1999), for example,
measured composite SWB scores. For example, Diener and all potential combinations of activated and deactivated posi-
Seligman (2002) used a composite measure of SWB formed tive and negative affective experiences are possible, with the
from the combination of various measures of LS, PA, and NA specific activation pattern depending on the situation and
to identify and compare participants in the top and bottom eliciting stimuli. Similarly, in the dynamic model of affect
10% of the distribution on the composite SWB score. Investi- (Reich, Zautra, & Davis, 2003), PA and NA are thought to be
gators have also examined SWB configurations using empiri- independently functioning systems and the relation between
cal means to identify individuals characterized by unique them is expected to become increasingly stronger and more
combinations of LS, PA, and NA. For example, Busseri, negative as stress levels increase.
Sadava, Molnar, and DeCourville (2009) used cluster analy- Despite these advantages, a configural structural concep-
sis to identity five distinct SWB configurations that replicated tualization requires several assumptions concerning the
across two large samples. As anticipated by Shmotkin (2005), underlying nature of SWB that are likely to be contentious.
the configurations identified by Busseri et al. included pat- As we have seen, research on SWB has focused almost
terns of high SWB (i.e., high LS, high PA, low NA) and low exclusively on relations involving LS, PA, and NA as dimen-
SWB, as well as various “incongruous” profiles, including a sional variables with interpretive emphasis given to individ-
profile of low LS despite moderate PA and NA (also see ual differences in rank ordering on each SWB component.
Bergman & Daukantaite, 2009). Identifying SWB configurations, in contrast, requires the
A configuration approach thus emphasizes the idiographic classification of individuals into different patterns based on
nature of SWB in which various SWB structures characterize distinct combinations of LS, PA, and NA. To date, research-
particular individuals rather than one structure applying to all ers have employed ad hoc cut scores and exploratory cluster
individuals. A configuration approach is also “person centered” analysis to identify distinct SWB configurations (e.g., Bergman
(e.g., Bergman & El-Khouri, 2003; Bergman, Magnusson, & & Daukantaite, 2009; Busseri et al., 2009; Shmotkin, 1998)
El-Khouri, 2001; Magnusson, 2003) in that emphasis is given rather than confirmatory techniques. Furthermore, ongoing
to organized patterns of SWB as an integrated system func- debates in other areas of psychological research concerning
tioning within individuals rather than simply to the individual the dimensional versus categorical nature of related phenom-
differences in LS, PA, and NA. From this perspective differ- ena such as personality traits (e.g., Asendorpf, 2003; Markon
ences in SWB configurations should have adaptive signifi- & Krueger, 2005) attest to the difficulty in establishing
cance such that knowledge concerning SWB should accrue unequivocally the utility of a person-centered approach.
from better understanding how individuals characterized by An additional limitation of the configural approach is
distinct configurations of LS, PA, and NA differ with respect that there is little clarity concerning which SWB configura-
to personality characteristics and life circumstances and in tions should be observed according to supporting concep-
various domains of functioning (Bergman & Daukantaite, tual models (e.g., Bergman & Daukantaite, 2009; Shmotkin,
2009; Busseri et al., 2009; Shmotkin et al., 2006). 2005). More generally, it is also uncertain which of the
various potential configurations best captures the notion of
Commentary. One advantage of treating SWB as a configuration SWB, or whether SWB refers to a general system of com-
of components is a supporting conceptual framework (Shmot- ponents rather than to a particular configuration of LS, PA,
kin, 2005) from which hypotheses concerning the structure and and NA.
function of SWB can be derived and tested empirically. A con-
figural conceptualization may also provide avenues for new
insights concerning, for example, the unique functional signifi- Commonalities and Discrepancies
cance of distinct SWB profiles and the possibility that a high Among Prominent Structural Models
SWB profile may not be the only configuration indicative of Despite the many differences among the five models of the
optimal human functioning. For example, according to Busseri tripartite structure of SWB that have been advanced over the
et al. (2009), individuals characterized by the combination of past four decades (see Table 1 for summary), there appears
moderate LS, low PA, and low NA reported levels of psycho- to be widespread agreement among SWB researchers con-
logical, physical, and interpersonal functioning that were com- cerning the importance of examining global life evaluations
parable in several ways to people characterized by the and affective reactions to one’s life in terms of LS, PA, and
combination of high LS, high PA, and low NA (i.e., high SWB). NA. Indeed, all five structural conceptualizations of SWB
Another advantage is that a configuration approach to the specify LS, PA, and NA as the primary components of SWB.
structure of SWB requires no assumptions concerning the This commonality provides an important starting point for

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


300 Personality and Social Psychology Review 15(3)

Table 1. Summary of the Five Prominent Models of the Tripartite Structure of Subjective Well-Being (SWB)

Model Description of SWB Emphasis Key strengths Key weaknesses


1. Separate Three separate components Separateness of LS, PA, Simplicity Status of SWB unclear
components (LS, PA, and NA) and NA as distinct
components
No assumptions Interpretation of
concerning associations correlations among
among components components unclear
2. Hierarchical Higher order factor Commonality among Informs nomological Requires negative
construct with three first-order LS, PA, and NA as networks of higher correlation between
indicators (LS, PA, and manifestations of latent order SWB and first- PA and NA
NA) SWB order components
Meaning of higher order
SWB unclear
Lack of evidence
concerning SWB as
common cause of LS,
PA, and NA
3.  Causal system LS is key criterion, Causal relations among Parsimony (single Referent for SWB
influenced by PA and NA components: PA and NA criterion) unclear
as inputs to LS
No constraint on PA–NA Lack of evidence
relation concerning causal
effect of PA and NA
on LS
4.  Composite Combination of three SWB as a composite score Informs nomological Meaning of associations
primary ingredients (LS, (i.e., LS + PA – NA) networks of composite among components
PA, and NA) SWB and ingredients unclear
No restrictions on Confusion with
associations among composite scores
components employed by
proponents of Model 2
5.  Configural Components (LS, PA, and Configurations or patterns Clear framework linking Lack of evidence
NA) organized in distinct of SWB components conceptualization and concerning categorical
ways within individuals operationalization nature of SWB
Flexibility concerning Number and nature
nature of associations of hypothesized
among components configurations unclear
Note: LS = life satisfaction; PA = positive affect; NA = negative affect.

comparison among structural models, as detailed below (see Emphasis of Separability Versus Commonality Among SWB
summary in Table 2). Components. Another important issue concerns the relative
emphasis on the separability versus commonality among
Status of SWB. A primary point of divergence among the promi- SWB components. The separate components approach
nent structural models concerns the best way to conceptualize (Model 1) focuses exclusively on the distinctiveness of LS,
SWB. In the separate components model (Model 1), SWB is PA, and NA. In contrast, both the hierarchical construct
considered to be a research domain and no causal connection (Model 2) and causal system (Model 3) conceptualizations
among components is implied. In the hierarchical construct emphasize the interconnections among SWB components,
conceptualization (Model 2), SWB is a higher order latent fac- although these latter two models differ in their respective
tor assumed to be responsible for producing the correlations focus on the common variance among versus directional
among components. In the causal system model (Model 3), relations between LS, PA, and NA. Between these extremes,
SWB refers to a network of relations in which PA and NA are the composite approach (Model 4) treats LS, PA, and NA as
thought to affect LS. In the composite conceptualization (Model separate causal indicators which, together, produce SWB,
4), SWB is created by the combination of LS, PA, and NA. In whereas the configuration approach (Model 5) emphasizes
the configuration model (Model 5), SWB is an integrated sys- the joint examination of within-individual combinations of
tem manifested within individuals as distinct configurations LS, PA, and NA to fully capture the organized, integrated
rather than a specific configuration of LS, PA, and NA. nature of the SWB system.

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 301

Table 2. Summary of Commonalities and Discrepancies Among Prominent Models of the Tripartite Structure of Subjective Well-Being
(SWB)

Separability vs. Meaning of shared


commonality of variance among Nature of PA–NA
Model Status of SWB components components relationship Role of components
1. Separate Research domain Separateness ? Not central Separate aspects of broad
components research domain
2. Hierarchical Hierarchical Commonality Underlying (latent) At least moderate in Reflective indicators of
construct (latent) SWB magnitude; negative higher order SWB; that
construct is, latent SWB causes
correlations among
components
3.  Causal system Network of Commonality Causal impact of PA, Not central PA and NA as causal
relations NA on LS predictors of LS (key
criterion)
4.  Composite Amalgam of LS, PA, ? ? Not central Causal indicators of a
and NA composite SWB; that is, LS,
PA, and NA produce SWB
5.  Configural Integrated system; ? ? Not central Separate aspects of
typology integrated system
Note:  LS = life satisfaction; PA = positive affect; NA = negative affect; ? = a lack of clarity in the model.

Interpretation of the Shared Variance Among SWB Components. composite, and configuration models (Models 1, 2, 4 and
A related source of disagreement pertains to the interpretation 5, respectively) each ascribe to LS, PA, and NA equivalent
of the shared versus unique variance among components. The conceptual roles (within a given model) as indicators of
hierarchical construct approach (Model 2) assumes that an SWB. However, the roles played by these three components
underlying sense of SWB is the cause of the positive correla- differ across models: separate aspects of a research domain
tions between LS and PA and the negative correlations (Model 1), reflective indicators of a higher order latent
between both of these components and NA. In contrast, under factor (Model 2), causal indicators of a latent composite
the causal system approach (Model 3), relations between LS variable (Model 4), and separate aspects of an integrated sys-
and PA and between LS and NA are interpreted as evidence tem (Model 5). In contrast, the causal system model (Model 3)
for the predictive roles of PA and NA as inputs to LS rather does not ascribe comparable roles to the three SWB compo-
than treating all three SWB components as indicators of the nents but rather delineates PA and NA as predictor variables
same underlying phenomenon. For the other three models and LS as the key criterion.
(Model 1, Model 4, and Model 5), the meaning of shared vari-
ance among SWB components is not clear. Dimensional Versus Categorical Nature of SWB. A final point of
divergence among structural models concerns the assumption of
Association Between PA and NA. A related issue concerns the an underlying dimensional versus categorical nature of SWB. In
magnitude and direction of the relation between PA and NA. the first four models, emphasis is given to associations involving
A substantive negative correlation between PA and NA is LS, PA, and NA as dimensional variables, each ranging from
critical to the viability of the hierarchical construct model low to high values. In contrast, the configuration approach
(Model 2). In contrast, in the separate components, causal sys- (Model 5) requires categorizing individuals based on unique pat-
tem, composite, and configural models (Models 1, 3, 4, and 5, terns of LS, PA, and NA rather than emphasizing individual dif-
respectively) the relation between PA and NA is not central to ferences in the level or amount on each dimension.
understanding SWB. Indeed, these models can be tested and
results interpreted regardless of the PA–NA relation. With
respect to the ongoing debate concerning the fundamental Implications
structure of affective experience, therefore, all of the models The emergence of five structural conceptualizations has
except Model 2 are viable regardless of whether PA and NA important implications for the study of SWB. As we describe
are orthogonal dimensions of affective experience or opposing below, improving our understanding of the structure of SWB
ends on a bipolar affect valence continuum. will have wide-ranging benefits with respect to measurement,
analysis and interpretation, and integration and synthesis.
Roles of LS, PA, and NA. The structural models also differ with
respect to the roles specified for the three components of Measurement. Diener and colleagues have repeatedly empha-
SWB. The separate components, hierarchical construct, sized the importance of measuring LS, PA, and NA in

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


302 Personality and Social Psychology Review 15(3)

studies examining SWB. The availability of well-studied, SWB. The separate components and causal system models
nonproprietary, and relatively brief self-report instruments (Model 1 and Model 4, respectively) require only three vari-
for assessing all three SWB components (e.g., Diener, ables—LS, PA, and NA. In contrast, the remaining three
Emmons, Larsen, & Griffin, 1985; Watson et al., 1988) structural approaches specify a fourth variable, including a
makes this recommendation feasible in a wide range of higher order factor (Model 2), a composite (Model 4), and a
research situations. Studies continue to be published, how- categorical marker of distinct configurations (Model 5). Thus,
ever, in which the authors claim to have examined SWB the required number of analysis variables differs across
while measuring only one component. The question can be structural conceptualizations.
raised, therefore, as to whether assessment of all three com- Furthermore, the most appropriate type of analysis will
ponents is necessary to advance our understanding of SWB. vary depending on the structural conceptualization one
In large part, the answer to this question depends on the adopts. The separate components model (Model 1) requires
researcher’s conceptualization of the structure of SWB. From researchers to examine the correlates, predictors, or potential
the perspective of the separate components model (Model 1), outcomes of LS, PA, and NA as separate outcomes. Such an
studying any of the three components of SWB, either together approach should also seek to control for the other two com-
or isolation, will advanced knowledge of SWB—particularly ponents (e.g., PA and NA) when determining relations involv-
if the term SWB is used to refer to a broad area of inquiry ing the third component (e.g., LS) to identify associations
rather than a psychological construct (e.g., Pavot, 2008). Simi- and isolate variance that is unique to each SWB component.
larly, proponents of the causal systems model (Model 3) might In contrast, a hierarchical conceptualization (Model 2) requires
argue that knowledge of SWB could accrue from studies researchers to employ an SEM approach and estimate SWB
examining LS even in isolation of PA and NA, given that LS as a latent factor, manifested in the common variance among
is sometimes considered to be the ultimate or key criterion of its three indicators. Correlates, predictors, and potential out-
interest in SWB research (e.g., Davern al., 2007; Oishi & Koo, comes of this latent SWB factor could then be explored
2008). With respect to the other three prominent structural along with simultaneous assessment of any residual links
models, however, assessment of all three SWB components is involving the individual SWB components independent of
required to inform SWB: Consistent with the hierarchical con- the higher order latent SWB factor. We urge proponents of a
struct model (Model 2), a latent SWB factor is estimated from hierarchical conceptualization to discontinue the use of a
the covariation among LS, PA, and NA; a composite SWB composite SWB index as a stand-in for a higher order latent
score (Model 4) requires all three components to be treated as SWB factor; as we have outlined above, an index score is
ingredients to the SWB; and the configuration approach consistent, both statistically and conceptually, with a com-
(Model 5) can be best informed by considering how LS, PA, posite conceptualization (Model 4) rather than a hierarchical
and NA are organized in unique ways within individuals. With construct model of SWB.
respect to these latter three structural models, therefore, SWB To continue, proponents of the causal systems model
cannot be studied without attention to all three components. (Model 3) need to employ a multiple regression, path analy-
Therefore, the first implication of the present review is that sis, or a SEM approach to test PA and NA as simultaneous
for at least three of the five tripartite structural models of SWB predictors of LS. Additional predictors (or correlates) of LS
assessing all three components (LS, PA, and NA) is neces- could then be explored after first accounting for the hypoth-
sary. Furthermore, given the current state of ambiguity con- esized proximal effects of PA and NA.
cerning the most appropriate structural model, we urge Further still, a basic approach to testing the composite
researchers interested in studying SWB by examining LS, PA, conceptualization (Model 4) would require researchers to
and NA as separate components or LS as the ultimate criterion compute an SWB index (e.g., an aggregate of standardized
of interest to address the implications of this decision, recog- scores for LS, PA, and reverse-scored NA), which is then
nizing that from the perspective of the other structural concep- explored in relation to other variables. Although less com-
tualizations, focusing on any individual component may not monly employed, a more sophisticated approach would be to
inform SWB. Indeed, we recommend that researchers, review- employ SEM to estimate SWB as a latent composite variable
ers, and editors discontinue publishing works nominally resulting from its three primary causal indicators. An impor-
examining SWB when, in fact, all three components have not tant advantage of this latter approach, compared to the use of
been addressed in the same study unless a compelling ratio- only a single SWB index, is that the correlates, predictors,
nale is provided for why this approach is appropriate and and potential outcomes of the latent composite SWB could
explicit consideration is given to what the implications may be be explored simultaneously with the assessment of potential
of focusing on individual SWB components. unique relations involving LS, PA, and NA.
Finally, researchers examining a configural conceptual-
Analysis and Interpretation. Even if one has collected data on ization of the structure of SWB (Model 5) need to employ
all three components of SWB, some structural conceptual- a person-centered statistical approach to identify and then
izations require the specification of a fourth variable to index compare among individuals characterized by distinct

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 303

configurations of LS, PA, and NA. Correlates, predictors, that may generalize across structural conceptualizations.
and potential outcomes predicted by SWB configurations Until such steps have been undertaken, we believe it prudent
could then be evaluated. to emphasize that researchers should not assume that find-
In light of these issues, the second major implication of ings based on one structural conceptualization of SWB apply
the present review is the need for careful consideration of and to alternative conceptualizations.
explicit justification for the analytic approach employed to
examine SWB-related data. We realize that given the current
state of ambiguity regarding the structure of SWB, determin- Steps Toward Deciding
ing the best analytic approach is a choice that researchers Among Structural Models
must make based on incomplete information. Regardless, Research addressing the disparities, inconsistencies, and
paralleling the need for researchers to explicitly identify the unanswered questions concerning the tripartite structure of
structural conceptualization employed in a given study, the SWB is overdue. To our knowledge, no systematic attempt
choice of analytic strategy also needs to be justified explicitly has been made to compare competing structural models of
rather than simply reported, and consideration should be SWB or to consider the conceptual and theoretical implica-
given to potential repercussions of the analytic choice. tions for understanding SWB more generally. Consequently,
the combination of compelling theoretical and empirical
Integration and Synthesis. A prerequisite for facilitating accurate arguments for adopting one structural conceptualization of
interpretation of SWB-related findings is unambiguous speci- SWB over the others has yet to be offered, and it is not clear
fication of which structural conceptualization of SWB is being that such a proclamation would be supportable at present
considered. Several nuanced and informative empirical given the ambiguities we have described. Similarly, in light
reviews of the characteristics, correlates, and predictors of of the numerous differences among structural conceptualiza-
SWB have been published (e.g., DeNeve & Cooper, 1999; tions, integration of the five models into a single structural
Diener, 1984; Diener et al., 1999; Howell & Howell, 2008; framework for SWB may not be realistic. Our view is that
Steel et al., 2008). However, these works have considered LS, SWB researchers instead need to evaluate the viability of all
PA, and NA separately—consistent with the separate compo- five models with the ultimate goal of deciding among the
nents model of the structure of SWB (Model 1). In contrast, various structural accounts based on a combination of con-
little is known about the correlates, predictors, and potential ceptual and empirical grounds. To these ends, there are sev-
consequences of SWB as a hierarchical construct based on eral valuable steps that should be given high priority.
reviews of research in which SWB is treated as a higher order
latent factor (Model 2), as a causal system based on reviews of Meta-Analysis. To date, hundreds of published empirical studies
research in which PA and NA are treated as predictors of LS have reported correlations among global self-reports of LS,
(Model 3), as a composite of three primary ingredients based PA, and NA. A meta-analysis of these correlations would pro-
on a review of research examining SWB as a composite index vide information central to the claims of several of the promi-
or latent composite variable (Model 4), or as configurations of nent SWB structural models. For example, consistent with the
components based on research examining SWB from a per- separate components model (Model 1) in which the relations
son-centered perspective (Model 5). among SWB components are not considered consequential,
Consequently, researchers employing one of these latter population estimates of the relations among all three SWB
four structural configurations (i.e., Models 2 through 5) nec- components may reveal insignificant or significant but small
essarily rely on reviews of the research literature treating correlations among LS, PA, and NA. It is also conceivable that
SWB as three separate components. However, it is not clear there are too many conditionalizing factors and too much vari-
how (and in what cases) the accumulated knowledge con- ability across studies to estimate a single, generalizable, and
cerning the nomological networks of LS, PA, and NA as reliable pattern of correlations among LS, PA, and NA (Diener,
separate components applies to a latent higher order SWB 1994, 2008; Pavot & Diener, 2008). Furthermore, with respect
factor, to a composite SWB index, to the causal links among to the viability of the hierarchical construct model (Model 2), if
PA, NA, and LS, or to SWB configurations. At present, the population estimates of the relations among global mea-
therefore, few integrative or summary statements can be sures of LS, PA, and NA are not substantive, the generalizabil-
offered about SWB as a higher order latent factor, causal ity of a higher order latent SWB factor would be called into
system, composite, or configuration of components. question. Similarly, regardless of the estimated PA–NA rela-
Given this situation, a third major implication of the pres- tion, population estimates for the LS–PA and LS–NA relations
ent review is that integration of SWB-related results may would also inform the viability of the causal system model
require separate reviews of different research literatures (Model 3) in which PA and NA are assumed to be significant
based on each of the five structural conceptualizations. Such positive and negative predictors of LS, respectively. Clearly,
steps will aid in the integration and synthesis of existing without both pairs of substantive relations, it would be hard to
model-specific knowledge as well as knowledge about SWB justify a structural model of SWB treating PA and NA as inputs

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


304 Personality and Social Psychology Review 15(3)

to LS. Thus, a meta-analysis would provide information criti- reveal important similarities and differences between con-
cal to understanding the structure of SWB, particularly with figurations not evident from treating SWB in a purely dimen-
respect to key assumptions in the separate components, hierar- sional manner. Thus, a compare and contrast approach will
chical construct, and causal system models. elucidate the extent to which varying the structural concep-
tualization leads to different (or similar) results and conclu-
Compare and Contrast. As researchers have demonstrated with sions concerning SWB.
respect to other multifaceted psychological phenomena (Carver, Furthermore, although empirical tests of the configuration
1989; Marsh & Yeung, 1998; Wallace & Bergeman, 2007), approach (Model 5) have employed cut scores or exploratory
adopting one structural conceptualization and accompanying cluster analyses to categorize individuals into distinct configu-
analytic strategy rather than a comparative approach may result ration groups, more advanced statistical techniques are avail-
in a loss of important information, create ambiguity concerning able that allow researchers to compare directly statistical
the nature of relations with other variables of theoretical inter- models composing dimensions versus categories (Muthén &
est, and lead to erroneous conclusions. In the present context, Muthén, 2000; Ruscio & Ruscio, 2008; Vermunt & Magidson,
some researchers have reported findings based on measured 2003). Consequently, as demonstrated in research on personal-
composite SWB scores or latent higher order SWB factors ity disorders (e.g., Krueger, Markon, Patrick, & Iacono, 2005),
alongside the individual SWB components (e.g., Romero, Vil- dimensional and categorical models could be compared using
lar, Luengo, & Gomez-Fraguela, 2009; Sheldon et al., 2004; latent trait and latent class approaches to provide a direct test
Vitterso, 2004), and one study has compared results based on a of a structural conceptualization for SWB composing catego-
configuration approach against findings derived from treating ries of SWB configurations versus SWB dimensions.
LS, PA, and NA as separate components (Bergman & Daukan- In the short term, applying these strategies would help
taite, 2009). We are aware of no previously published studies, raise awareness among SWB researchers concerning the
however, in which all five structural models of SWB have been implications of adopting any particular structural conceptu-
contrasted with respect to associations and predictive relations alization of SWB. As an immediate step toward this goal, the
with other variables. Consequently, it is unclear the extent to abundant data sets that have already been reported in studies
which results based on one structural approach and accompany- of SWB in which all three components have been assessed
ing analytic strategy would be similar to findings determined could be profitably reconsidered by comparing and contrast-
based on any of the other conceptualizations. For example, it is ing the five structural models of SWB. In the longer term, the
uncertain whether the five structural models would produce accumulation of findings from comparisons among alterna-
similar or differing results concerning the correlates, predictors, tive structural models based on both existing and newly col-
and potential consequences of SWB. lected data will provide valuable insights concerning the
Yet an important consequence of the commonality among advantages and disadvantages of one structural conceptual-
structural conceptualizations in delineating LS, PA, and NA ization of SWB over the others.
as the components of SWB is that data from a study based on
one structural approach can be used to evaluate each of the Longitudinal Studies. Another method for evaluating the five
other structural models, assuming that all three SWB compo- prominent structural models of SWB is to conduct longitudinal
nents are assessed. For example, researchers examining the studies in which LS, PA, and NA are tracked over repeated
association between a factor such as income and SWB could assessments. Reports of large-scale long-term panel studies
evaluate the claim from the separate components model typically focus on LS (e.g., Lucas & Donnellan, 2007; Luh-
(Model 1) that LS, PA, and NA are distinct components with mann & Eid, 2009) rather than all three SWB components.
discriminable sets of associations with other variables. This Even in studies examining longitudinal relations among LS,
claim could be tested directly against the claim that SWB is PA, and NA (e.g., Busseri et al., 2008; Lyubomirsky et al.,
most appropriately structured as a hierarchical construct 2005; Molnar et al., 2009), comparative evaluations among
(Model 2) by determining whether any of the individual structural conceptualization have not been undertaken. The
components have unique relations with income, independent existence of such data sets, however, raises the possibility that
of a latent higher order SWB factor. Consistent with the researchers could reexamine these data and comparatively eval-
causal system approach (Model 3), PA and NA could also be uate the five structural models in a longitudinal context. In addi-
evaluated as potential mediating factors of the effect of tion to informing the viability of each structural conceptualization,
income on LS. In addition, a latent composite SWB variable the use of longitudinal designs would allow for important tests
(Model 4) could be estimated as a product of LS, PA, and concerning relations between SWB and other variables.
NA, and relations could be evaluated between income and For example, consistent with the separate components
the latent composite SWB variable as well as the unique model (Model 1), prospective linkages among SWB compo-
variance in each of the individual SWB components. Finally, nents (e.g., LS at one time point predicting changes in NA at
SWB configurations (Model 5) could be identified and a subsequent assessment) should not be substantive. Further-
examined with respect to differences in income, which may more, researchers should be able to identify prospective

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 305

associations involving other variables of interest (e.g., per- PA, or NA should be mediated by the higher order latent
sonality traits, demographic factors) that are unique to each SWB factor. Alternatively, consistent with the causal system
of the three SWB components. In contrast, according to the model (Model 3), interventions designed to change PA or
hierarchical construct model (Model 2), a latent higher order NA should lead indirectly to changes in LS. It may also be
SWB factor should predict changes in LS, PA, and NA over possible to design interventions to manipulate LS indepen-
time, independent of stability unique to the individual com- dent of PA and NA, thereby informing LS-specific (vs.
ponents. In addition, predictive effects of other variables affect-mediated) causal relations involving other theoreti-
should be strongest on the latent SWB factor, and prospec- cally informative factors (e.g., Schimmack et al., 2008).
tive effects on LS, PA, and NA should be indirect via effects A composite conceptualization (Model 4) predicts that
on the latent SWB factor (e.g., Molnar et al., 2009). Alterna- induced changes to any one SWB component should also
tively, according to the causal system model (Model 3), both lead to changes in the latent composite SWB variable, even
PA and NA should predict changes in LS over time, rather if changes in the other SWB components were not observed.
than vice versa. Furthermore, PA and NA could be tested as In addition, consistent with the assumption that SWB is
mediators to determine whether prospective effects of other manifest as a combination of components, it should be pos-
factors (e.g., extraversion and neuroticism) on changes in LS sible to identify factors that, when manipulated experimen-
are mediated by or independent of PA and NA (see Klonowicz, tally, have their greatest impact on the composite SWB
2001; Schimmack et al., 2002; Urry et al., 2004). variable compared to the resulting impact on any of the indi-
From the perspective of the composite model (Model 4), vidual causal indicators. Finally, according to the configura-
changes in the latent composite SWB variable should be esti- tion approach (Model 5), induced changes in one SWB
mated through the combination of changes in its three causal component should affect the entire SWB system, potentially
indicators. In addition, prospective relations of other factors on resulting in the reorganization of one’s SWB components.
the latent SWB composite should be indirect over time, via The nature of this potential reorganization may depend on
changes in all or some of the three causal indicators. Finally, to both the type of manipulation employed (i.e., the SWB com-
evaluate the configuration approach (Model 5), both sample- ponent targeted) and the person’s SWB configuration prior
level structural stability and intraindividual stability in SWB to the experiment. Thus, although relatively rarely employed
configuration over time could be determined. Furthermore, pro- in SWB research, experimental approaches could be profit-
spective connections between SWB configurations and other ably employed to address issues concerning the structure
indicators of functioning could be evaluated to assess whether of SWB.
SWB configurations respond to changes in life circumstances
and adversity (Shmotkin, 2005). In summary, the opportunity to
evaluate several issues fundamental to each of the structural Summary and Conclusion
models makes the analysis of longitudinal SWB data, including A tripartite account of SWB encompasses global life evalua-
reanalysis of existing data sets, an important future direction. tions and positive and negative affective experiences. Over
the past four decades, researchers have adopted various con-
Experimental Studies. Complementing the naturalistic study ceptualizations concerning the structure of SWB with respect
of variability in SWB via longitudinal designs, valuable to its primary components: LS, PA, and NA. We have
information concerning the internal structure of SWB could reviewed five structural models found in contemporary
also come from experimental studies. Interest in SWB- research on SWB. One model treats SWB as three separate
related interventions is growing, and a variety of potentially components and emphasizes the relative distinctiveness of
effective techniques to alter levels of LS, PA, and NA have LS, PA, and NA (Model 1). A second approach conceptual-
been reported (see Seligman, Steen, Park, & Peterson, 2005; izes SWB as a hierarchical construct composing a higher
Sin & Lyubomirsky, 2009). What is needed, however, are order SWB factor and LS, PA, and NA as three first-order
investigations tracking participants over long enough peri- indicators (Model 2). A third structural conceptualization
ods of time to monitor both short-term and longer-term specifies a causal system in which LS, the key criterion of
impacts on all three SWB components (see Okun, Olding, & interest, is directly influenced by PA and NA (Model 3). A
Cohn, 1990) and to evaluate observed impacts from the per- fourth model casts the structure of SWB in terms of a com-
spective of all five structural models. posite, resulting from the combination of LS, PA, and NA
According to the separate components model (Model 1), (Model 4). A fifth conceptualization (Model 5) emphasizes
researchers should be able to independently manipulate each SWB as an organized system in which LS, PA, and NA are
SWB component. The hierarchical construct model (Model 2), configured in distinctive ways within individuals.
in contrast, predicts that induced changes in LS, PA, or NA These five models share the assumption that LS, PA, and
resulting from a change in participants’ underlying sense of NA compose the core components of SWB. However, little
well-being may also lead to changes in the other components. consensus exists concerning how these components should
Furthermore, attempts to manipulate change in SWB via LS, be treated. Rather, structural models are characterized by

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


306 Personality and Social Psychology Review 15(3)

conflicting assumptions concerning several fundamental issue. We anticipate that an immediate benefit of undertak-
issues, including the status of SWB as a psychological con- ing any of these steps will be heightened awareness among
struct versus domain of inquiry, the causal connections SWB researchers of the existence of multiple diverging
between SWB and its components, the importance of the sep- structural perspectives on SWB and greater sensitivity to the
arateness versus commonality among components, the mean- importance of addressing unresolved discrepancies and
ing of the relations among SWB components, the nature of ambiguities concerning the structure of SWB.
the PA–NA relation, the roles played by each SWB compo- In conclusion, the tripartite structure of SWB has yet to be
nent, and the dimensional versus categorical nature of SWB. determined. Over two decades ago Diener (1984) observed
The commonality and differences among structural that the relations among SWB components had not been
accounts have important implications. First, from the per- thoroughly researched and thus recommended “a continuous
spective of at least three of the five structural models, mea- interchange between data and theoretical positions” (p. 569).
surement of LS, PA, and NA within the same study is crucial Although hundreds of studies are now published on SWB
to build a knowledge base concerning SWB based on joint each year, Diener’s advice appears to have gone largely
consideration of all three components. Second, models differ unheeded with respect to the structure of SWB. Although it
in the number of analysis variables required to study SWB is premature to propose a definitive structure for SWB,
(i.e., three components vs. three components plus a separate improving our understanding of the structure of SWB will
SWB variable). Furthermore, because the most appropriate have wide-ranging benefits. Greater clarity concerning
analytic approach for SWB-related data differs according to whether and (if so) how LS, PA, and NA fit together will
the presumed structural model, and given the absence of increase the likelihood that the database of SWB-related
clarity at present concerning the “correct” structure of SWB, research findings can be synthesized accurately and effec-
researchers should clearly specify the underlying structural tively. More immediate, illuminating the structure of SWB
model assumed in a given study and justify their decisions will help SWB researchers address practical issues concern-
concerning analytic approach. Third, with respect to integra- ing analysis and interpretation of SWB-related findings.
tion and synthesis, whereas existing reviews of SWB-related Ultimately, therefore, resolving the structure of SWB is cru-
research primarily inform the separate components model in cial to our understanding of SWB and what it means to expe-
which LS, PA, and NA are treated as distinct components, rience and evaluate life in positive ways.
whether these reviews inform the correlates, predictors, or
consequences of SWB as a hierarchical construct, a compos- Declaration of Conflicting Interests
ite or causal system, or distinct configurations has not been The author(s) declared no potential conflicts of interests with respect
determined. Consequently, researchers should not assume to the authorship and/or publication of this article.
that the current SWB knowledge base necessarily applies to
each structural conceptualization. Financial Disclosure/Funding
Given these implications, improvement in our under- The author(s) received no financial support for the research and/or
standing of the structure of SWB will have wide-ranging authorship of this article.
benefits. Toward this goal, we have described a broad agenda
for research on the structure of SWB, constituting both short- Notes
term and longer-term strategies. Several steps were delin- 1. Consonant with its subjective focus, subjective well-being
eated, including meta-analyzing the relations among LS, PA, (SWB) is measured almost exclusively using self-report ratings
and NA; systematically and empirically comparing the five of life satisfaction (LS), positive affect (PA), and negative affect
models; tracking SWB components longitudinally over time; (NA). Furthermore, although Diener’s (1984) tripartite model
and employing experimental studies designed to manipulate was subsequently expanded to encompass domain-specific sat-
components of SWB to observe changes in the other compo- isfaction judgments, such as satisfaction with one’s job or mar-
nents and, perhaps, the relations among (or configurations riage (e.g., Diener, 2000; Diener, Suh, Lucas, & Smith, 1999;
created by) these components. No single study addressing Diener, Tamir, & Napa Scollon, 2006), the typical study of SWB
any of these goals will provide definitive answers. Rather, examines all three components at a global level, based on the
each of these steps will require a targeted program of research overall evaluation of one’s life and positive and negative affec-
involving both the collection of new SWB-related data as tive experiences in general (Diener, 2008; Schimmack, 2008).
well as the reanalysis of existing data sets. Among the earliest SWB measures was a single-item self-
Importantly, although we hope that the present review anchoring ladder developed by Kilpatrick and Cantril (1960).
will stimulate new research, we are not simply calling for This approach and other single-item measures of global life eval-
“additional research” to be added to the fast-growing moun- uations (e.g., “How satisfied are you with your life, overall?”)
tain of empirical work on SWB. Rather, we wish to call are still widely employed. Another popular LS measure is the
attention to the need for systematic investigation of the struc- multi-item Satisfaction With Life Scale, developed by Diener,
ture of SWB given the foundational implications of this Emmons, Larsen, and Griffin (1985), which comprises five

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 307

items concerning overall evaluations of one’s life (e.g., “The between SWB and happiness, however, has been specified in
conditions of my life are excellent”). With respect to global various ways. For example, some researchers have promoted
affective reactions, the most widely used measure is the Positive the position that global happiness is synonymous with SWB
and Negative Affect Schedule (PANAS), developed by Watson, (e.g., Biswas-Diener, Vitterso, & Diener, 2005; Diener &
Clark, and Tellegen (1988), which comprises 10 items assessing Fujita, 1995; Diener, Sandvik, & Pavot, 1991; Diener, Sand-
a general tendency toward positive affectivity and 10 items vik, Pavot, & Gallagher, 1991; Diener, Sandvik, Seidlitz,
assessing a tendency toward negative affectivity. Although both & Diener, 1993; Gohm, Oishi, Darlington, & Diener, 1998;
state and trait versions of the PANAS have been employed, the Larsen, Diener, & Emmons, 1985; Pavot & Diener, 1993;
trait measure, in which respondents indicate their affective Pavot, Diener, Colvin, & Sandvik, 1991). This view is conso-
experiences “in general,” is most commonly used (Schimmack nant with other research in which global happiness is consid-
& Crites, 2005). ered to be interchangeable with a higher order latent well-being
2. Although a tripartite model of SWB has served as the founda- factor (e.g., Stones & Kozma, 1980, 1985). In such accounts,
tion for a large volume of research, other major models of well- happiness and SWB are assumed to “share a common core of
being also have been proposed. Waterman (1993) differentiated meaning” (Stones & Kozma, 1989, p. 526; also see Larsen &
between hedonic well-being, which pertains to enjoyment and McKibban, 2008; van Praag, Frijters, & Ferrier-i-Carbonell,
satisfaction, and eudaimonic well-being, which addresses self- 2003). In other models, happiness is specified as one of sev-
actualization and personal growth (also see Ryan & Deci, 2001; eral lower order indicators of a higher order SWB factor (e.g.,
Ryff, 1989). Although there have been debates concerning the Arthaud-Day, Rode, Mooney, & Near, 2005; Kozma, Stone,
most appropriate model of well-being (e.g., Diener, Sapyta, Stones, Hannah, & McNeil, 1990; McKennell, 1978; McNeil,
& Suh, 1998; Ryff & Singer, 1998), researchers have consis- Stones, & Kozma, 1986; Oishi & Koo, 2008; Schyns, 1998).
tently found strong positive correlations between the two types Diener and colleagues have also equated happiness with the
of well-being (e.g., Compton, 2001; Keyes, Shmotkin, & Ryff, affective components of SWB only (e.g., Diener & Diener, 1996;
2002; King, Hicks, Krull, & Del Gaiso, 2006; van Dierendonck, Seidlitz & Diener, 1993), consonant with Bradburn’s (1969) in-
2005; Vitterso, 2003). An emerging body of research seeks to fluential early formulation (also see Stones & Kozma, 1986).
understand this relation between hedonic and eudaimonic well- Other researchers have conceptualized happiness as a person-
being (Diener, Oishi, & Lucas, 2003; Keyes, 1998; King & ality trait that is more general than, and thus subsumes, SWB
Napa, 1998; Seligman, Steen, Park, & Peterson, 2005; Westerhof, (e.g., Lyubomirsky, 2001; Lyubomirsky & Lepper, 1999). Fur-
Dittman-Kohli, & Thissen, 2001). Clearly, SWB is not the only ther still, it has also been proposed that global measures of hap-
conceptualization of interest to well-being researchers, nor does piness “cannot be neatly mapped onto the components of SWB”
it provide a complete account of positive quality of life (Keyes, (Diener et al., 1999, p. 277). At present, therefore, the rela-
2003; Keyes et al., 2002). An important priority for future tion between SWB and happiness is unclear.
research is to develop a “more elaborate, yet more precise, under-
standing of the interrelations of SWB with other faculties of References
well-being” (Shmotkin, Berkovich, & Cohen, 2006, p. 140). Adler, M. G., & Fagley, N. S. (2005). Appreciation: Individual dif-
3. Consistent with early formulations (e.g., Bradburn, 1969; Costa & ferences in finding value and meaning as a unique predictor of
McCrae, 1980), researchers sometimes combine PA and NA into subjective well-being. Journal of Personality, 73, 79–114.
an “affect balance” score (i.e., PA minus NA), also referred to as Anderson, C. A., & Sedikides, C. (1991). Thinking about people:
“hedonic level” or “hedonic balance” (Basabe et al., 2002; Chwal- Contributions of a typological alternative to associationistc and
isz, Diener, & Gallagher, 1988; Diener, 1994; Diener & Biswas- dimensional models of person perception. Journal of Personal-
Diener, 2002; Diener, Sandvik, Pavot, & Gallagher, 1991; Diener ity and Social Psychology, 60, 203–217.
et al., 1993; Diener, Lucas, Oishi, & Suh, 2002; Diener, Sandvik, Andrews, F. M., & Crandall, R. (1976). The validity of measures of
Pavot, & Fujita, 1992; Diener, Wolsic, & Fujita, 1995; Larsen self-reported well-being. Social Indicators Research, 3, 1–19.
et al., 1985; Pavot & Diener, 1993; Pavot et al., 1990; Pavot et Andrews, F. M., & McKennell, A. C. (1980). Measures of self-
al., 1997; Schimmack et al., 2002; Suh, Diener, Oishi, & Tri- reported well-being: Their affective, cognitive, and other com-
andis, 1998; Tov & Diener, 2007). In other reports, however, ponents. Social Indicators Research, 8, 127–155.
researchers have stressed that measures of PA and NA should Andrews, F. M., & Robinson, J. P. (1991). Measures of subjective
not be combined into a single score in light of the loss of valu- well-being. In J. P. Robinson, P. R. Shaver, & L. S. Wrightman
able information concerning the two types of affect (e.g., Die- (Eds.), Measures of personality and social psychological atti-
ner, 2000; Diener & Suh, 1999). In the present work, we fo- tudes (pp. 61–114). San Diego, CA: Academic Press.
cus on the structure of SWB with respect to LS, PA, and NA Andrews, F. M., & Withey, S. B. (1976). Social influences of well-
rather than based on the combination of PA and NA into an being. New York, NY: Plenum.
affect balance score. Argyle, M., & Martin, M. (1991). The psychological causes of hap-
4. Diener (1984) suggested that SWB is typically what people piness. In F. Strack, M. Argyle, & N. Schwarz (Eds.), Subjective
have in mind when they refer to “happiness.” The relationship well-being (pp. 77–100). London, England: Pergamon.

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


308 Personality and Social Psychology Review 15(3)

Arthaud-Day, M. L., Rode, J. C., Mooney, C. H., & Near, J. P. Busseri, M. A., Sadava, S. W., Molnar, D. S., & DeCourville, N.
(2005). The subjective well-being construct: A test of its con- (2009). A person-centered approach to subjective well-being.
vergent, discriminant, and factorial validity. Social Indicators Journal of Happiness Studies, 10, 161–181.
Research, 74, 445–476. Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1999). The
Asendorpf, J. B. (2003). Head-to-head comparison of the predictive affect system has parallel and integrative processing compo-
validity of personality types and dimensions. European Journal nents: Form follows function. Journal of Personality and Social
of Personality, 17, 327–346. Psychology, 76, 839–855.
Ashby, F. G., Isen, A. M., & Turken, A. U. (1999). A neuropsycho- Campbell, A. (1976). Subjective measures of well-being. American
logical theory of positive affect and its influence on cognition. Psychologist, 31, 117–124.
Psychological Review, 106, 529–550. Campbell, A. (1981). The sense of well-being in America. New
Balatsky, G., & Diener, E. (1993). Subjective well-being among York, NY: McGraw-Hill.
Russian students. Social Indicators Research, 28, 225–243. Campbell, A., Converse, P. E., & Rodgers, W. L. (1976). The qual-
Basabe, N., Paez, D., Valencia, J., Gonzalez, J. L., Rime, B., & ity of American life: Perceptions, evaluations, and satisfaction.
Diener, E. (2002). Cultural dimensions, socioeconomic devel- New York, NY: Russell Sage.
opment, climate, and emotional hedonic level. Cognition and Carlisle, S., & Hanlon, P. (2007). Well-being and consumer cul-
Emotion, 16, 103–125. ture: A different kind of public health problem? Health Promo-
Beiser, M. (1974). Components and correlates of mental well- tion International, 22, 261–268.
being. Journal of Health and Social Behavior, 15, 320–327. Carp, F. M., & Carp, A. C. (1982). Test of a model of domain satis-
Bentler, P. M., & Kano, Y. (1990). On the equivalence of factors factions and well-being. Research on Aging, 4, 503–522.
and components. Multivariate Behavioral Research, 25, 67–74. Carp, F. M., & Carp, A. C. (1983). Structural stability of well-being
Bergman, L. R., & Daukantaite, D. (2009). Stability of typical pat- factors across age an gender, and development of scales of well-
terns of subjective well-being in middle-aged Swedish women. being unbiased for age and gender. Journal of Gerontology, 38,
Journal of Happiness Studies, 10, 293–311. 572–581.
Bergman, L. R., & El-Khouri, B. M. (2003). A person-oriented Carver, C. S. (1989). How should multifaceted personality con-
approach: Methods for today and methods for tomorrow. New structs be studied? Issues illustrated by self-monitoring, attri-
Directions in Child and Adolescent Development, 101, 25–38. butional style, and hardiness. Journal of Personality and Social
Bergman, L. R., Magnusson, D., & El-Khouri, B. M. (2001). Studying Psychology, 56, 577–585.
individual development in an interindividual context. Mahwah, Chamberlain, J. (1988). On the structure of subjective well-being.
NJ: Erlbaum. Social Indicators Research, 20, 581–604.
Biswas-Diener, R., Vitterso, J., & Diener, E. (2005). Most people Cheng, S.-T. (2004). Age and subjective well-being revisited: A
are pretty happy, but there is cultural variation: The Inughuit, the discrepancy perspective. Psychology and Aging, 19, 409–415.
Amish, and the Maasai. Journal of Happiness Studies, 6, 205–226. Chwalisz, K., Diener, E., & Gallagher, D. (1988). Autonomic
Bollen, K. (2002). Latent variables in psychology and the social arousal feedback and emotional experience: Evidence from the
sciences. Annual Review of Psychology, 53, 605–634. spinal cord injured. Journal of Personality and Social Psychol-
Bollen, K., & Lennox, R. (1991). Conventional wisdom on mea- ogy, 54, 820–828.
surement: A structural equation perspective. Psychological Bul- Compton, W. C. (2001). Toward a tripartite factor structure of men-
letin, 110, 305–314. tal health: Subjective well-being, personal growth, and religios-
Bood, S.-A., Archer, T., Norlander, T. (2004). Affective personal- ity. Journal of Psychology, 135, 486–500.
ity in relation to general personality, self-reported stress, cop- Costa, P. T., & McCrae, R. R. (1980). Influence of extraversion
ing, and optimism. Individual Differences Research, 2, 26–37. and neuroticism on subjective well-being: Happy and unhappy
Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2003). The people. Journal of Personality and Social Psychology, 38,
theoretical status of latent variables. Psychological Review, 110, 668–678.
203–219. Cummins, R. A., & Nistico, H. (2002). Maintaining life satisfac-
Bradburn, N. M. (1969). The structure of psychological well-being. tion: The role of positive cognitive bias. Journal of Happiness
Oxford, England: Aldine. Studies, 3, 37–69.
Bradburn, N. M., & Caplovitz, D. (1965). Reports on happiness. Davern, M., & Cummins, R. A. (2006). Is life dissatisfaction the
Chicago, IL: Aldine. opposite of life satisfaction? Australian Journal of Psychology,
Brenner, B. (1975). Quality of affect and self-evaluated happiness. 58, 1–7.
Social Indicators Research, 2, 315–331. Davern, M., Cummins, R. A., & Stokes, M. A. (2007). Subjective
Busseri, M. A., Sadava, S. W., & DeCourville, N. (2007). A hybrid wellbeing as an affective-cognitive construct. Journal of Happi-
model for research on subjective well-being: Examining common- ness Studies, 8, 429–449.
and component-specific sources of variance in life satisfaction, DeNeve, K. M., & Cooper, H. (1998). The happy personality: A
positive affect, and negative affect. Social Indicators Research, meta-analysis of 137 personality traits and subjective well-
83, 413–445. being. Psychological Bulletin, 124, 197–229.

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 309

Diener, E. (1984). Subjective well-being. Psychological Bulletin, life satisfaction judgments. Personality and Social Psychology
95, 542–575. Bulletin, 28, 437–445.
Diener, E. (1994). Assessing subjective well-being: Progress and Diener, E., Napa Scollon, C. K., & Lucas, R. E. (2003). The evolv-
opportunities. Social Indicators Research, 31, 103–157. ing concept of subjective well-being: The multifaceted nature
Diener, E. (1995). A value based index for measuring national qual- of happiness. Advances in Cell Aging and Gerontology, 15,
ity of life. Social Indicators Research, 36, 107–127. 187–219.
Diener, E. (1996). Traits can be powerful, but are not enough: Les- Diener, E., Napa Scollon, C. K., Oishi, S., Dzokoto, V., & Suh, E. M.
sons from subjective well-being. Journal of Research in Per- (2000). Positivity and the construction of life satisfaction judg-
sonality, 30, 389–399. ments: Global happiness is not the sum of its parts. Journal of
Diener, E. (2000). Subjective well-being: The science of happiness Happiness Studies, 1, 159–176.
and a proposal for a national index. American Psychologist, 55, Diener, E., Oishi, S., & Lucas, R. E. (2003). Personality, culture,
34–43. and subjective well-being: Emotional and cognitive evaluations
Diener, E. (2008). Myths in the science of happiness, and direc- of life. Annual Review of Psychology, 54, 403–425.
tions for future research. In M. Eid & R. J. Larsen (Eds.), The Diener, E., Sandvik, E., & Pavot, W. (1991). Happiness is the fre-
science of subjective well-being (pp. 493–514). New York, NY: quency, not the intensity, of positive versus negative affect. In
Guilford. F. Strack, M. Argyle, & N. Schwarz (Eds.), Subjective well-
Diener, E., & Biswas-Diener, R. (2002). Will money increase sub- being (pp. 119–139). London, England: Pergamon.
jective well-being? Social Indicators Research, 57, 119–169. Diener, E., Sandvik, E., Pavot, W., & Fujita, F. (1992). Extraver-
Diener, E., & Diener, C. (1996). Most people are happy. Psycho- sion and subjective well-being in a U.S. national probability
logical Science, 7, 181–185. sample. Journal of Research in Personality, 26, 205–215.
Diener, E., Diener, M., & Diener, C. (1995). Factors predicting the Diener, E., Sandvik, E., Pavot, W., & Gallagher, D. (1991).
subjective well-being of nations. Journal of Personality and Response artifacts in the measurement of subjective well-being.
Social Psychology, 69, 851–864. Social Indicators Research, 24, 35–56.
Diener, E., & Emmons, R. A. (1985). Personality correlates of sub- Diener, E., Sandvik, E., Seidlitz, L., & Diener, M. (1993). The rela-
jective well-being. Personality and Social Psychology Bulletin, tionship between income and subjective well-being: Relative or
11, 89–97. absolute? Social Indicators Research, 28, 195–223.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Diener, E., Sapyta, J. J., & Suh, E. (1998). Subjective well-being is
Satisfaction with Life Scale. Journal of Personality Assessment, essential to well-being. Psychological Inquiry, 9, 33–37.
49, 71–75. Diener, E., & Seligman, M. E. P. (2002). Very happy people. Psy-
Diener, E., Gohm, C. L., Suh, E., & Oishi, S. (2000). Similarity of the chological Science, 13, 81–84.
relations btween martial status and subjective well-being across Diener, E., & Suh, E. M. (1999). National differences in subjective
cultures. Journal of Cross-Cultural Psychology, 31, 419-436. well-being. In D. Kahneman, E. Diener, & N. Schwarz (Eds.),
Diener, E., Kesebir, P., & Lucas, R. (2008). Benefits of accounts of Well-being: The foundations of hedonic psychology (pp. 434–450).
well-being: For societies and for psychology science. Applied New York, NY: Russell Sage.
Psychology: An International Review, 57, 37–53. Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Sub-
Diener, E., Larsen, R. J., Levine, S., & Emmons, R. A. (1985). jective well-being: Three decades of progress. Psychological
Intensity and frequency: Dimensions underlying positive and Bulletin, 125, 276–302.
negative affect. Journal of Personality and Social Psychology, Diener, E., Suh, E., & Oishi, S. (1997). Recent findings on subjective
48, 1253–1265. well-being. Indian Journal of Clinical Psychology, 24, 25–41.
Diener, E., & Lucas, R. E. (1999). Personality and subjective well- Diener, E., Suh, E., Smith, H., & Shao, L. (1995). National differ-
being. In Kahneman, Diener, & Schwartz (Eds.). Well-being: ences in reported subjective well-being: Why do they occur?
The foundation of hedonic psychology (pp.213–229). New York, Social Indicators Research, 34, 7–32.
NY: Russell Sage. Diener, E., Tamir, M., & Napa Scollon, C. (2006). Happiness, life
Diener, E., & Lucas, R. E. (2000a). Explaining differences in soci- satisfaction, and fulfillment: The social psychology of subjective
etal levels of happiness: Relative standards, need fulfillment, well-being. In P. A. M. Van Lange (Ed.), Bridging social psychol-
culture, and evaluation theory. Journal of Happiness Studies, ogy: Benefits of transdisciplinary approaches (pp. 319–324).
1, 41–78. Mahwah, NJ: Erlbaum.
Diener, E., & Lucas, R. E. (2000b). Subjective emotional well- Diener, E., & Tov, W. (2007). Subjective well-being and peace.
being. In M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of Journal of Social Issues, 63, 421–440.
emotions (pp. 325–337). New York, NY: Guilford. Diener, E., Wolsic, B., & Fujita, F. (1995). Physical attractiveness
Diener, E., Lucas, R. E., & Napa Scollon, C. (2006). Beyond the and subjective well-being. Journal of Personality and Social
hedonic treadmill: Revising the adaptation theory of well-being. Psychology, 69, 120–129.
American Psychologist, 61, 305–314. Dolan, P., & White, M. (2006). Dynamic well-being: Connecting
Diener, E., Lucas, R. E., Oishi, S., & Suh, E. M. (2002). Looking indicators of what people anticipate with indicators of what they
up and looking down: Weighing good and bad information in experience. Social Indicators Research, 75, 303–333.

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


310 Personality and Social Psychology Review 15(3)

Edwards, J. R., & Bagozzi, R. P. (2000). On the nature and direc- Keyes, C. L. M. (2003). Complete mental health: An agenda for
tion of relationships between constructs and measures. Psycho- the 21st century. In C. L. M. Keyes & J. Haidt (Eds.), Flourish-
logical Methods, 5, 155–174. ing: Positive psychology and the life well-lived (pp. 293–312).
Eid, M. (1997). Happiness and satisfaction: An application of Washington, DC: American Psychological Association.
a latent state-trait model for ordinal variables. In J. Rost & Keyes, C. L. M. (2005). Mental health and/or mental illness? Inves-
R. Langehein-Munster (Eds.), Applications of latent trait and tigating axioms of the complete state model of health. Journal of
latent class models in the social sciences (pp. 148–154). New Consulting and Clinical Psychology, 73, 539–548.
York, NY: Waxmann. Keyes, C. L. M., Shmotkin, D., & Ryff, C. D. (2002). Opti-
Eid, M., & Larsen, R. J. (2008). The science of subjective well- mizing well-being: The empirical encounter of two tradi-
being. New York, NY: Guilford. tions. Journal of Personality and Social Psychology, 82,
Elliot, A. J., Gable, S. L., & Mapes, R. R. (2006). Approach and 1007–1022.
avoidance motivation in the social domain. Personality and Kilpatrick, F. P., & Cantril, H. (1960). Self-anchoring scaling: A
Social Psychology Bulletin, 32, 378–391. measure of individuals’ unique reality worlds. Journal of Indi-
Emmons, R. A., & Diener, E. (1985). Personality correlates of sub- vidual Psychology, 16, 158–173.
jective well-being. Personality and Social Psychology Bulletin, Kim-Prieto, C., Diener, E., Tamir, M., Scollon, C., & Diener, M. (2005).
11, 89–97. Integrating the diverse definitions of happiness: A time-sequential
Fredrickson, B. L. (2001). The role of positive emotions in posi- framework of subjective well-being. Journal of Happiness Studies,
tive psychology: The broaden-and-build theory of positive emo- 6, 261–300.
tions. American Psychologist, 56, 218–226. King, L. A., Hicks, J. A., Krull, J. L., & Del Gaiso, A. K. (2006).
George, J. M. (1991). Time structure and purpose as a mediator of Positive affect and the experience of meaning in life. Journal of
work-life linkages. Journal of Applied Social Psychology, 21, Personality and Social Psychology, 90, 179–196.
296–314. King, L. A., & Napa, C. S. (1998). What makes life good? Journal
George, L. K. (1981). Subjective well-being: Conceptual and meth- of Personality and Social Psychology, 75, 156–165.
odological issues. Annual Review of Gerontology and Geriat- Kline, R. B. (1998). Principles and practice of structural equation
rics, 2, 345–382. modeling. New York, NY: Guilford.
Gohm, C. L., Oishi, S., Darlington, J., & Diener, E. (1998). Cul- Klonowicz, T. (2001). Discontented people: Reactivity and locus
ture, parental conflict, parental marital status, and the subjective of control as determinants of subjective well-being. European
well-being of young adults. Journal of Marriage and Family, 60, Journal of Personality, 15, 29–47.
319–334. Kozma, A., & Stone, S. (1980). The measurement of happiness:
Grice, J. W. (2001). Computing and evaluating factor scores. Psy- Development of the Memorial University of Newfoundland
chological Methods, 6, 430–450. Scale of Happiness (MUNSH). Journal of Gerontology, 35,
Heller, D., Komar, J., & Lee, W. B. (2007). The dynamics of per- 906–912.
sonality states, goals, and well-being. Personality and Social Kozma, A., Stone, S., & Stones, M. J. (2000). Stability in compo-
Psychology Bulletin, 33, 898–910. nents and predictors of subjective well-being (SWB): Implica-
Heller, D., Watson, D., & Ilies, R. (2006). The role of person versus tions for SWB structure. In E. Diener & D. R. Rahtz (Eds.),
situation in life satisfaction: A critical examination. Psychologi- Advances in quality of life theory and research (pp. 13–30).
cal Bulletin, 130, 574–600. London, England: Kluwer.
Ho, S. M. Y., & Cheng, M. W. L. (2007). Using the combined Kozma, A. Stone, S., Stones, M. J., Hannah, T. E., & McNeil, K.
etic-emic approach to develop a measurement of interpersonal (1990). Long- and short-term affective states in happiness:
subjective well-being in Chinese populations. In A. D. Ong & Model, paradigm and experimental evidence. Social Indicators
M. H. van Dulmen (Eds.), Oxford handbook of methods in posi- Research, 22, 119–138.
tive psychology (pp. 139–152). New York, NY: Oxford Univer- Krueger, R. F., Markon, K. E., Patrick, C. J., & Iacono, W. G. (2005).
sity Press. Externalizing psychopathology in adulthood: A dimensional-
Howell, R. T., & Howell, C. J. (2008). The relation of economic spectrum conceptualization and its implications for DSM-V.
status to subjective well-being in developing countries: A meta- Journal of Abnormal Psychology, 114, 537–550.
analysis. Psychological Bulletin, 134, 536–560. Kuppens, P., Realo, A., & Diener, E. (2008). The role of positive and
Jarvis, C. B., MacKenzie, S. B., & Podsakoff, P. M. (2003). A criti- negative emotions in life satisfaction judgments across nations.
cal review of construct indicators and measurement model mis- Journal of Personality and Social Psychology, 95, 66–75.
specification in marketing and consumer research. Journal of Larsen, R. J., Diener, E., & Emmons, R. A. (1985). An evaluation
Consumer Research, 30, 199–218. of subjective well-being measures. Social Indicators Research,
Kammann, R., Farry, M., & Herbison, P. (1984). The analysis and 17, 1–17.
measurement of happiness as a sense of well-being. Social Indi- Larsen, R. J., & Eid, M. (2008). Ed Diener and the science of
cators Research, 15, 91–115. subjective well-being. In M. Eid & R. J. Larsen (Eds.), The
Keyes, C. L. M. (1998). Social well-being. Social Psychology science of subjective well-being (pp. 1–13). New York, NY:
Quarterly, 61, 121–140. Guilford.

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 311

Larsen, R. J., & McKibban, A. R. (2008). Is happiness having what Lynn, M., & Steel, P. (2006). National differences in subjective
you want, wanting what you have, or both? Psychological Sci- well-being: The interactive effects of extraversion and neuroti-
ence, 19, 371–377. cism. Journal of Happiness Studies, 7, 155–165.
Larsen, R. J., & Prizmic, Z. (2008). Regulation of emotional well- Lyubomirsky, S. (2001). Why are some people happier than others?
being: Overcoming the hedonic treadmill. In M. Eid & R. J. Larsen American Psychologist, 56, 239–249.
(Eds.), The science of subjective well-being (pp. 258–289). Lyubomirsky, S., & Lepper, H. S. (1999). A measure of subjec-
New York, NY: Guilford. tive happiness: Preliminary reliability and construct validation.
Lawrence, R. H., & Liang, J. (1988). Structural integration of the Social Indicators Research, 46, 137–155.
Affect Balance Scale and the Life Satisfaction Scale Index A: Race, Lyubomirsky, S., Sheldon, K. M., & Schkade, D. (2005). Pursu-
sex, and age differences. Psychology and Aging, 3, 375–384. ing happiness: The architecture of sustainable change. Review
Lawton, M. P. (1982). The well-being and mental health of of General Psychology, 9, 111–131.
the aged. In T. M. Field, A. Huston, H. C. Quay, L. Troll, & MacCallum, R. C., & Browne, M. W. (1993). The use of causal
G. E. Finley (Eds.), Review of human development (pp. 614–628). indicators in covariance structure models: Some practical issues.
New York, NY: Wiley. Psychological Bulletin, 114, 533–541.
Lawton, M. P. (1983a). Environment and other determinants of MacKenzie, S. B., Podsakoff, P. M., & Jarvis, C. B. (2005). The
well-being in older people. Gerontologist, 23, 349–357. problem of measurement model misspecification in behavioral
Lawton, M. P. (1983b). The varieties of wellbeing. Experimental and organizational research and some recommended solutions.
Aging Research, 9, 65–72. Journal of Applied Psychology, 90, 710–730.
Lent, R. W. (2004). Toward a unifying theoretical and practical per- Magnusson, D. (2003). The person-centered approach: Concepts,
spective on well-being and psychological adjustment. Journal measurement models, and research strategy. New Directions in
of Counseling Psychology, 51, 482–509. Child and Adolescent Development, 101, 3–23.
Liang, J. (1984). Dimensions of the Life Satisfaction Index A: A Malka, A., & Chatman, J. A. (2003). Intrinsic and extrinsic orienta-
structural formulation. Journal of Gerontology, 39, 613–622. tions as moderators of the effect of annual income on subjective
Liang, J. (1985). A structural integration of the Affect Balance well-being: A longitudinal study. Personality and Social Psy-
Scale and the Life Satisfaction Index A. Journal of Gerontol- chology Bulletin, 29, 737–746.
ogy, 40, 552–561. Markon, K. E., & Krueger, R. F. (2005). Categorical and continuous
Liang, J., & Bollen, K. A. (1983). The structure of the Philadelphia models of liability to externalizing disorders: A direct comparison
Geriatric Center Morale Scale: A reinterpretation. Journal of in NESARC. Archives of General Psychiatry, 62, 1352–1359.
Gerontology, 38, 181–189. Marsh, H. W., & Yeung, A. S. (1998). Top-down, bottom-up, and
Liang, J., & Bollen, K. A. (1985). Sex differences in the structure horizontal models: The direction of causality in multidimen-
of the Philadelphia Geriatric Center Morale Scale. Journal of sional hierarchical self-concepts. Journal of Personality and
Gerontology, 40, 468–477. Social Psychology, 75, 509–527.
Linley, P. A., Maltby, J., Wood, A. M., Osborne, G., & Hurling, R. McCrae, R. M., & John, O. P. (1992). An introduction to the five-
(2009). Measuring happiness: The higher order factor structure factor model and its applications. Journal of Personality, 60,
of subjective and psychological well-being measures. Personal- 175–215.
ity and Individual Differences, 47, 878–884. McCulloch, B. J. (1991). A longitudinal investigation of the factor
Lohmann, N. (1977). Correlations of life satisfaction, morale and structure of subjective well-being: The case of the Philadelphia
adjustment measures. Journal of Gerontology, 32, 73–75. Geriatric Center Morale Scale. Journal of Gerontology: Psycho-
Lucas, R. E. (2007). Using structural equation models to validate logical Sciences, 46, P251–P258.
measures of positive emotions. In A. D. Ong & M. H. van Dulmen McCullough, G., Huebner, E. S., & Laughlin, J. E. (2000). Life
(Eds.), Oxford handbook of methods in positive psychology events, self-concept, and adolescents’ positive subjective well-
(pp. 111–125). New York, NY: Oxford University Press. being. Psychology in the Schools, 37, 281–290.
Lucas, R. E. (2008). Personality and subjective well-being. In M. McKennell, A. C. (1978). Cognitive and affect in perceptions of
Eid & R. J. Larsen (Eds.), The science of subjective well-being well-being. Social Indicators Research, 5, 389–426.
(pp. 171–194). New York, NY: Guilford. McKennell, A. C., & Andrews, F. M. (1980). Models of cogni-
Lucas, R. E., Diener, E., & Suh, E. (1996). Discriminant validity tion and affect in perceptions of well-being. Social Indicators
of well-being measures. Journal of Personality and Social Psy- Research, 8, 257–298.
chology, 71, 616–628. McNeil, J. K., Stones, M. J., & Kozma, A. (1986). Subjective well-
Lucas, R. E., & Donnellan, M. B. (2007). How stable is happiness? being in later life: Issues concerning measurement and predic-
Using the STARTS model to estimate the stability of life sat- tion. Social Indicators Research, 18, 35–70.
isfaction. Journal of Research in Personality, 41, 1091–1098. Michalos, A. C. (1980). Satisfaction and happiness. Social Indica-
Luhmann, L., & Eid, M. (2009). Does it really feel the same? tors Research, 8, 385–422.
Changes in life satisfaction following repeated life events. Jour- Molnar, D., Busseri, M. A., Perrier, C. P., & Sadava, S. (2009).
nal of Personality and Social Psychology, 97, 363–381. A longitudinal investigation of alcohol use and subjective

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


312 Personality and Social Psychology Review 15(3)

well-being in a university sample. Journal of Studies on Alcohol cultural identification and language. Psychological Science,
and Drugs, 70, 704–713. 18, 607–613.
Muthén, B. O., & Muthén, L. K. (2000). Integrating person- Pressman, S. D., & Cohen, S. (2005). Does positive affect influence
centered and variable-centered analyses: Growth mixture mod- health? Psychological Bulletin, 131, 925–971.
eling with latent trajectory classes. Alcoholism: Clinical and Reich, J. W., Zautra, A. J., & Davis, M. (2003). Dimensions of
Experimental Research, 24, 882–891. affect relationships: Models and their integrative implications.
Napier, J. L., & Jost, J. T. (2008). Why are conservatives happier Review of General Psychology, 7, 66–83.
than liberals? Psychological Science, 19, 565–572. Robinson, M. D., Crawford Solberg, E., Vargas, P. T., & Tamir, M.
Nes, R. B., Roysamb, E., Tambs, K., Harris, J. R., & Reichborn- (2003). Trait as default: Extraversion, subjective well-being,
Kjennerud, T. (2006). Subjective well-being: Genetic and envi- and the distinction between neutral and positive faces. Journal
ronmental contributions to stability and change. Psychological of Personality and Social Psychology, 85, 517–527.
Medicine, 36, 1033–1042. Romero, E., Villar, P., Luengo, M. A., & Gomez-Fraguela, J. A.
Norlander, T., Bood, S.-A., & Archer, T. (2002). Performance dur- (2009). Traits, personal strivings, and well-being. Journal of
ing stress: Affective personality, age, and regularity of physical Research in Personality, 43, 535–546.
exercise. Social Behavior and Personality, 30, 495–508. Ruscio, J., & Ruscio, A. M. (2008). Categories and dimensions:
Oishi, S., & Diener, E. (2001). Re-examining the general positiv- Advancing psychological science through the study of latent
ity model of subjective well-being: The discrepancy between structure. Current Directions in Psychological Science, 17,
specific and global domain satisfaction. Journal of Personality, 203–207.
69, 641–666. Russell, J. A., & Barrett, L. F. (1999). Core affect, prototypical emo-
Oishi, S., Diener, E., & Lucas, R. E. (2007). The optimum level of tional episodes, and other things called emotion: Dissecting the ele-
well-being: Can people be too happy? Perspectives on Psycho- phant. Journal of Personality and Social Psychology, 76, 805–819.
logical Science, 2, 346–360. Ryan, R. M., & Deci, E. L. (2001). On happiness and human poten-
Oishi, S., & Koo, M. (2008). Two new questions about happiness: tials: A review of research on hedonic and eudaimonic well-
“Is happiness good?” and “Is happiness better?” In M. Eid & being. Annual Review of Psychology, 52, 141–166.
R. J. Larsen (Eds.), The science of subjective well-being (pp. Ryff, C. D. (1989). Happiness is everything, or is it? Explorations
290–306). New York, NY: Guilford. on the meaning of psychological well-being. Journal of Person-
Okun, M. A., Olding, R. W., & Cohn, C. M. (1990). A meta-analysis ality and Social Psychology, 57, 1069–1081.
of subjective well-being interventions among elders. Psycholog- Ryff, C. D., & Singer, B. (1998). The contours of positive human
ical Bulletin, 108, 257–166. health. Psychological Inquiry, 9, 1–28.
Pavot, W. (2008). The assessment of subjective well-being: Suc- Sagiv, L., & Schwartz, S. H. (2000). Value priorities and subjective
cesses and shortfalls. In M. Eid & R. J. Larsen (Eds.), The sci- well-being: Direct relations and congruity of effects. European
ence of subjective well-being (pp. 124–140). New York, NY: Journal of Social Psychology, 30, 1177–198.
Guilford. Sandvik, E., Diener, E., & Seidlitz, L. (1993). Subjective well-
Pavot, W., & Diener, E. (1993). The affective and cognitive context being: The convergence and stability of self-report and non-self-
of self-reported measures of subjective well-being. Social Indi- report measures. Journal of Personality, 61, 317–342.
cators Research, 28, 1–20. Schimmack, U. (2008). The structure of subjective well-being. In
Pavot, W., & Diener, E. (2004). Findings on subjective well-being: M. Eid & R. J. Larsen (Eds.), The science of subjective well-
Applications to public policy, clinical interventions, and educa- being (pp. 97–123). New York, NY: Guilford.
tion. In A. P. Linely & S. Joseph (Eds.), Positive psychology in Schimmack, U., & Crites, S. L. (2005). The structure of affect. In
practice (pp. 679–692). Hoboken, NJ: Wiley. D. Albarracin, B. T. Johnson, & M. P. Zanna (Eds.), The hand-
Pavot, W., & Diener, E. (2008). The Satisfaction With Life Scale book of attitudes (pp. 397–435). Mahwah, NJ: Erlbaum.
and the emerging construct of life satisfaction. Journal of Posi- Schimmack, U., Diener, E., & Oishi, S. (2002). Life-satisfaction is
tive Psychology, 3, 137–152. a momentary judgment and a stable personality characteristic:
Pavot, W., Diener, E., Colvin, C. R., & Sandvik, E. (1991). Further The use of chronically accessible and stable sources. Journal of
validation of the Satisfaction With Life Scale: Evidence of the Personality, 70, 345–384.
cross-method convergence of well-being measures. Journal of Schimmack, U., Schupp, J., & Wagner, G. G. (2008). The influ-
Personality Assessment, 57, 149–161. ence of environment and personality on the affective and cog-
Pavot, W., Diener, E., & Fujita, F. (1990). Extraversion and hap- nitive components of subjective well-being. Social Indicators
piness. Personality and Individual Differences, 11, 1299–1306. Research, 89, 41–60.
Pavot, W., Fujita, F., & Diener, E. (1997). The relation between Schyns, P. (1998). Crossnational differences in happiness: Eco-
self-aspect congruence, personality, and subjective well-being. nomic and cultural factors explained. Social Indicators Research,
Personality and Individual Differences, 22, 183–191. 43, 3–26.
Perunovic, W. Q. E., Heller, D., & Rafaeli, E. (2007). Within- Scollon, C. N., Diener, E., Oishi, S., & Biswas-Diener, R. (2005).
person changes in the structure of emotion: The role of An experience sampling and cross-cultural investigation of the

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


Busseri and Sadava 313

relation between pleasant and unpleasant affect. Cognitive and Steiger, J. H. (1990). Some additional thoughts on components,
Emotion, 19, 27–52. factors, and factor indeterminacy. Multivariate Behavioral
Seidlitz, L., & Diener, E. (1993). Memory for positive versus nega- Research, 25, 41–45.
tive life events: Theories fsor the differences between happy and Stones, M. J., & Kozma, A. (1980). Issues relation to the usage and
unhappy persons. Journal of Personality and Social Psychol- conceptualization of mental health constructs employed by ger-
ogy, 64, 654–664. ontologist. International Journal of Aging and Human Develop-
Seidlitz, L., Wyer, R. S., & Diener, E. (1997). Cognitive correlates ment, 11, 269–281.
of subjective well-being: The processing of valenced life events Stones, M. J., & Kozma, A. (1985). Structural relationships among
by happy and unhappy persons. Journal of Research in Person- happiness scales: A second order factorial study. Social Indica-
ality, 31, 240–256. tors Research, 17, 19–28.
Seligman, M. E. P. (2000). Positive psychology: An introduction. Stones, M. J., & Kozma, A. (1986). “Happy are they who are
American Psychologist, 55, 5–14. happy . . .” A test between two causal models of relationships
Seligman, M. E. P., Steen, T., Park, N., & Peterson, C. (2005). Posi- between happiness and its correlates. Experimental Aging
tive psychology progress: Empirical validation of interventions. Research, 12, 23–39.
American Psychologist, 60, 410–421. Stones, M. J., & Kozma, A. (1989). Happiness and activities in later
Sheldon, K. M., Elliot, A. J., Ryan, R. M., Chirkov, V., Yougmee, K., life: A propensity formulation. Canadian Psychology, 30, 526–537.
Ku, C.,  .  .  .  Sun, Z. (2004). Self-concordance and subjective Storbeck, J., & Clore, G. L. (2008). The affective regulation of cog-
well-being in four cultures. Journal of Cross-Cultural Psychol- nitive priming. Emotion, 8, 208–215.
ogy, 35, 209–223. Suh, E. M. (2002). Culture, identity consistency, and subjective
Sheldon, K. M., & Hoon, T. H. (2007). The multiple determina- well-being. Journal of Personality and Social Psychology, 83,
tion of well-being: Independent effects of positive traits, needs, 1378–1391.
goals, selves, social supports, and cultural contexts. Journal of Suh, E., Diener, E., Oishi, S., & Triandis, H. C. (1998). The shifting
Happiness Studies, 8, 565–592. basis of life satisfaction judgments across cultures: Emotions
Sheldon, K. M., & Lyubomirsky, S. (2006). Achieving sustain- versus norms. Journal of Personality and Social Psychology,
able gains in happiness: Change your actions, not your circum- 74, 482–493.
stances. Journal of Happiness Studies, 7, 55–86. Tellegen, A. (1985). Structures of mood and personality and their
Shmotkin, D. (1998). Declarative and differential aspects of subjec- relevance to assessing anxiety, with an emphasis on self-report.
tive well-being and its implications for mental health in later In T. A. Hussain & J. D. Maser (Eds.), Anxiety and the anxiety
life. In J. Lomranz (Ed.), Handbook of aging and mental health: disorders (pp. 681–706). Hillsdale, NJ: Erlbaum.
An integrative approach (pp. 15–43). New York, NY: Plenum. Tov, W., & Diener, E. (2007). Culture and subjective well-being. In
Shmotkin, D. (2005). Happiness in the face of adversity: Reformu- S. Kitayama & D. Cohen (Eds.), Handbook of cultural psychol-
lating the dynamic and modular bases of subjective well-being. ogy (pp. 691–713). New York, NY: Guilford.
Review of General Psychology, 9, 291–325. Urry, H. L., Nitschke, J. B., Dolski, I., Jackson, D. C., Dalton, K. M.,
Shmotkin, D., Berkovich, M., & Cohen, K. (2006). Combining hap- Mueller, C. J., . . . Davidson, R. J. (2004). Making a life worth
piness and suffering in a retrospective view of anchor periods living: Neural correlates of well-being. Psychological Science,
in life: A differential approach to subjective well-being. Social 15, 367–372.
Indicators Research, 77, 139–169. van Dierendonck, D. (2005). The construct validity of Ryff’s Scales
Shmotkin, D., & Hadari, G. (1996). An outlook on subjective well- of Psychological Well-being and its extension with spiritual
being in older Israeli adults: A unified formulation. Interna- well-being. Personality and Individual Differences, 36, 629–
tional Journal of Human Development, 42, 271–289. 643.
Sin, N. L., & Lyubomirsky, S. (2009). Enhancing well-being and van Praag, B. M. S., Frijters, P., & Ferrier-i-Carbonell, A. (2003).
alleviating depressive symptoms with positive psychology inter- The anatomy of subjective well-being. Journal of Economic
ventions: A practice-friendly meta-analysis. Journal of Clinical Behavior and Organization, 51, 29–49.
Psychology, 65, 467–487. Vermunt, J. K., & Magidson, J. (2003). Latent class models for
Singh, K., & Jha, S. D. (2008). Positive and negative affect, and grit classification. Computational Statistics and Data Analysis, 41,
as predictors of happiness and life satisfaction. Journal of the 541–537.
Indian Academy of Applied Psychology, 34, 40–45. Vitterso, J. (2001). Personality traits and subjective well-being:
Sirgy, M. J., Michalos, A. C., Ferriss, A. L., Easterlin, R. A., Emotional stability, not extraversion, is probably the impor-
Patrick, D., & Pavot, W. (2006). The quality-of-life (QOL) tant predictor. Personality and Individual Differences, 31,
research movement: Past, present, and future. Social Indicators 903–914.
Research, 76, 343–466. Vitterso, J. (2003). Flow versus life satisfaction: A projective use
Steel, P., Schmidt, J., & Shultz, J. (2008). Refining the relationship of cartoons to illustrate the difference between the evaluation
between personality and subjective well-being. Psychological approach and the intrinsic motivation approach to subjective
Bulletin, 134, 138–161. well-being. Journal of Happiness Studies, 4, 141–167.

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014


314 Personality and Social Psychology Review 15(3)

Vitterso, J. (2004). Subjective well-being versus self-actualization: Watson, D., Weise, D., Vaidya, J., & Tellegen, A. (1999). The two
Using the flow-simplex to promote a conceptual clarification general activation systems of affect: Structural findings, evolu-
of subjective quality of life. Social Indicators Research, 65, tionary considerations, and psychobiological evidence. Journal
299–331. of Personality and Social Psychology, 76, 820–838.
Vitterso, J., & Nilsen, F. (2002). The conceptual and relational Westerhof, G. J., Dittman-Kohli, F., & Thissen, T. (2001). Beyond
structure of subjective well-being, neuroticism, and extraver- life satisfaction: Lay conceptions of well-being among mid-
sion. Social Indicators Research, 57, 89–118. dle-aged and elderly adults. Social Indicators Research, 56,
Wallace, K. A., & Bergeman, C. S. (2007). Hardiness as a dispo- 179–203.
sitional resource: Methods of conceptualizing the construct. In Westerhof, G. J., & Keyes, C. L. M. (2006). After the fall of the
A. D. Ong & M. H. van Dulmen (Eds.), Oxford handbook of Berlin Wall: Perceptions and consequences of stability and change
methods in positive psychology (pp. 323–336). New York, NY: among middle-aged and older East and West Germans. Journal
Oxford University Press. of Gerontology: Social Sciences, 61B, S240–S247.
Waterman, A. S. (1993). Two conceptions of happiness: Contrasts Westerhof, G. J., Thissen, T., Dittman-Kohli, F., & Stevens, N. L.
of personal expressiveness (eudaimonia) and hedonic enjoyment. (2006). What is the problem? A taxonomy of life problems and
Journal of Personality and Social Psychology, 64, 678–691. their relation to subjective well-being in middle and late adult-
Watkins, P. C., Woodward, K., Stone, T., & Kolts, R. L. (2003). hood. Social Indicators Research, 79, 91–115.
Gratitude and happiness: Development of a measure of gratitude Wiese, B. S. (2007). Successful pursuit of personal goals and sub-
and relationships with subjective well-being. Social Behavior jective well-being. In B. R. Little, K. Salmela-Aro, & S. D.
and Personality, 31, 431–452. Phillips (Eds.), Personal project pursuit: Goals, action and
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and human flourishing (pp. 301–328). Hillsdale, NJ: Erlbaum.
validation of brief measures of positive and negative affect: The Zelenski, J. M., Murphy, S. A., & Jenkins, D. A. (2008). The
PANAS scales. Journal of Personality and Social Psychology, happy-productive worker thesis revisited. Journal of Happiness
54, 1063–1070. Studies, 9, 521–537.

Downloaded from psr.sagepub.com at University of Birmingham on November 20, 2014

You might also like