Content uploaded by Sonja Lyubomirsky
Author content
All content in this area was uploaded by Sonja Lyubomirsky on Nov 13, 2017
Content may be subject to copyright.
KENNON M. SHELDON and SONJA LYUBOMIRSKY
ACHIEVING SUSTAINABLE GAINS IN HAPPINESS:
CHANGE YOUR ACTIONS, NOT YOUR
CIRCUMSTANCES
w
ABSTRACT. Although attaining happiness is a nearly universal goal, sur-
prisingly little research has focused on how happiness can be increased and
then sustained. Three studies test predictions of a model (Lyubomirsky et al.,
2005) positing that sustainable happiness is possible through intentional
activity changes, more so than through circumstantial changes. Study 1 shows
that less hedonic adaptation is reported in response to activity changes than to
circumstantial changes. Study 2 tests a dynamic process model, showing that
while both positive activity changes and positive circumstantial changes predict
rank-order increases in subjective well-being from Time 1 to Time 2, only
activity changes predict maintained gains at Time 3. Study 3 replicates the
Study 2 findings and extends them to psychological well-being (Ryff and
Keyes, 1995). Implications for positive psychology and ‘‘the pursuit of
happiness’’ are discussed.
KEY WORDS: happiness, hedonic adaptation, set point, well-being.
THE ‘‘PURSUIT OF HAPPINESS’’
The ‘‘pursuit of happiness’’ is an American cultural obsession,
beginning with the Declaration of Independence, where it is
pledged as an important right for all citizens. Today, happiness
remains a topic of tremendous interest for groups as diverse as
philosophers, policy makers, and poets, and, increasingly, for
economists, popular psychology writers, and happiness ‘‘coa-
ches.’’ As any bookstore visitor can attest, rows of self-help
books in any major bookstore are committed to the literature on
happiness. Furthermore, the pursuit of happiness is becoming
w
This work was supported in part by grants from the Positive Psychology
Network.
Journal of Happiness Studies (2006) 7:5586 ÓSpringer 2006
DOI 10.1007/s10902-005-0868-8
ever more global, as people seek to realize the promises of capi-
talism and political freedom (Freedman, 1978; Diener et al.,
1995).
Reflecting this burgeoning interest, research on subjective
well-being (SWB) has exploded over the last 20 years, with hun-
dreds of citations per year. A variety of correlates of happiness,
positive mood, and life satisfaction have now been identified
(DeNeve and Cooper, 1998). For example, well-being has been
shown to be associated with certain demographic variables (e.g.,
Argyle, 1999; Diener et al., 1999), personality traits and atti-
tudes (e.g., Diener and Lucas, 1999), and goal characteristics
(e.g., McGregor and Little, 1998).
Ironically, however, very few of these studies speak to a cen-
tral assumption of the ‘‘American dream’’ namely, the idea
that one can achieve and maintain relatively greater well-being,
compared to an earlier period in one’s life. At least three waves
of data are required to test this assumption, but the vast major-
ity of studies of well-being have used data from only one time
period. In other words, most studies have been cross-sectional
rather than longitudinal. Thus, researchers still have little
understanding of how well-being varies over time, and what fac-
tors influence these changes (but see Atkinson, 1982, for an ear-
ly discussion of this issue, or more recently, Lucas et al., 2003).
In addition, little is known about what factors, if any, might
bring about stable changes in levels of well-being, either positive
or negative.
Reasons for Pessimism Regarding Sustainable Increases
in Well-Being
Doubtless, part of the reason researchers have neglected to
study sustainable well-being change is the difficulty of conduct-
ing longitudinal and prospective studies. However, we believe
another reason for this neglect is the considerable scientific pes-
simism over whether it is even possible to achieve sustainable in-
creases in happiness. One source of pessimism is the idea of a
genetically-determined set point for happiness. Lykken and Telle-
gen (1996) have provided evidence, based on behavior-genetic
studies, that the heritability of well-being may be as high as
80% (although 50% is a more widely accepted figure; Diener
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
56
et al., 1999). Regardless of the exact coefficient, the large
magnitude of these estimates suggests that each person has a
chronic or characteristic level of happiness, from which it will
be difficult, if not impossible, to depart. Thus, although there
may be substantial variation around this baseline level in the
short-term, in the long-term, people perhaps cannot help but re-
turn to their set point: ‘‘What goes up must come down’’ (e.g.,
Headey and Wearing, 1989).
A second source of pessimism comes from the concept of he-
donic adaptation (Brickman and Campbell, 1971; Frederick and
Loewenstein, 1999), which suggests that gains in happiness are
impermanent, because humans so quickly adapt to change (see
also Kahneman, 1999; Scitovsky, 1976). That is, although new
circumstances may cause temporary increases in happiness or
sadness, people rapidly adjust, and the effect of these new cir-
cumstances on their well-being then quickly diminishes or even
entirely disappears. For example, Brickman et al. (1978) showed
that recent lottery winners were no happier than controls and,
furthermore, that recent victims of paralysis were not as
unhappy as one would expect (see also Dijkers, 1997).
Reasons for Optimism Regarding Sustainable Increases
in Well-Being
Do the arguments for pessimism described above lead to the
conclusion that the pursuit of happiness is fruitless that, ra-
ther than chasing happiness, people may be better off by simply
accepting their current personality and happiness levels
(McCrae and Costa, 1994)? For three reasons, we believe not.
First, some researchers have had preliminary success, albeit
short-term, in using interventions to enhance happiness (e.g.,
Schulz, 1976; Fordyce, 1977, 1983; Lichter et al., 1980; Fava,
1999; Sheldon et al., 2002). The potential of happiness-increas-
ing interventions is further demonstrated by recent research
showing that practicing certain virtues, such as gratitude (Em-
mons and McCullough, 2003), forgiveness (McCullough et al.,
2000), and thoughtful self-reflection (King, 2001; Lyubomirsky
et al., 2004) can bring about enhanced well-being.
Second, many different motivational factors have been
associated with well-being, factors that are presumably amenable
SUSTAINABLE HAPPINESS 57
to some volitional control. Examples of motivational factors in-
clude the successful pursuit of life goals that are intrinsic in con-
tent (e.g., Kasser and Ryan, 1993, 1996), concordant with an
individual’s interests, values, and motives (Brunstein et al., 1998;
Sheldon and Elliot, 1999), and internally consistent (e.g., Emmons
and King, 1988; Sheldon and Kasser, 1995). Thus, one might
find greater happiness by choosing particular kinds of goals.
Third, a variety of cognitive factors have been linked to well-
being, factors that are presumably also amenable to some voli-
tional control. These include pausing to count one’s blessings
(Emmons and McCullough, 2003), eschewing social compari-
sons and contingent self-evaluations (e.g., Lyubomirsky and
Ross, 1997), and choosing to feel a sense of optimism or effi-
cacy regarding one’s life (Taylor and Brown, 1988; Seligman,
1991; Bandura, 1997). Thus, by changing one’s patterns of
thought and ways of construing events, one might find greater
happiness.
To summarize, there appears to be a paradox: Some theoreti-
cal perspectives and empirical data imply that well-being can be
increased, while other theories and data suggest that it cannot.
How can these conflicting perspectives on the possibility of last-
ing happiness change be resolved? Also, if increases in happi-
ness are indeed possible, what kinds of circumstances, actions,
or habits of mind are most likely to bring gains, especially gains
that can be maintained?
A New Model of Longitudinal Well-Being
In this article, we test important predictions from a new model
of longitudinal well-being recently advanced by Lyubomirsky
et al. (2005). The model specifies three major determinants of
well-being at time t: (1) the person’s genetic set point or set
range (which reflects personality and temperament), (2) the per-
son’s current circumstances (demographic, geographic, and con-
textual), and (3) the person’s current intentional activities
(behavioral, cognitive, and conative). We focus on these three
major categories because they have historically received the
majority of attention in the well-being literature (Diener et al.,
1999), and because they allow consideration of several important
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
58
issues and paradoxes, such as the question of whether it is even
possible to ‘‘become happier’’ given strong genetic influences on
happiness, the question of why past well-being research has
found such weak associations between demographic/circumstantial
variables and happiness, and the question of how a person might
appropriately take action to ‘‘pursue’’ happiness.
We assume that the set point remains constant across the
lifespan, reflecting the person’s basic temperament, constitution,
and personality traits. Because the factors that determine the set
point (i.e., the person’s basic temperament and personality
traits) are, by definition, stable, they should have little or no
impact on variations in well-being over time. Thus, the set point
is not formally measured or modeled in the current research.
Instead, we assume that it is best estimated by the person’s
average well-being score (Lykken, 1999) and, furthermore, that
it accounts for the considerable expected longitudinal stability
in well-being. Although some personality measures may tap the
set point relatively directly, we assumed that including such
constant measures in our models would not affect the dynamic
patterns observed.
What then accounts for variations in an individual’s level of
well-being? Our model specifies that the person’s current circum-
stances (e.g., his health, income, or the region where he lives)
can either add to or detract from his constant set point, as can
the person’s current intentional activities (e.g., her behavioral
activity of exercising regularly, her cognitive activity of trying to
accentuate the positive, or her conative activity of trying to gain
admission to a professional program). As Lykken (2000) has ar-
gued, despite a genetically-determined baseline for well-being,
humans are capable of increasing their happiness relative to this
baseline through various ‘‘happiness makers.’’ In this article, we
subdivide these happiness makers into two types activity-
based and circumstance-based.
As evident from the above descriptions, activity-based
changes, by definition, involve continual effort and engagement
in some intentional process, whereas circumstance-based
changes are, by definition, one-time changes that tend to occur
independently of effort and engagement. Of course, the distinc-
tion between activity and circumstantial changes may not
SUSTAINABLE HAPPINESS 59
always be clear, as activity may be required to change one’s
circumstances, and because many circumstances involve activity.
Despite these challenges, we attempted to separately measure
these two abstract categories, hoping to illustrate with psycho-
metric and hypothesis-confirmatory results that they are valid
and important.
A central assumption of Lyubomirsky et al.’s (2005) model is
that hedonic adaptation occurs more quickly with respect to cir-
cumstantial changes than to activity changes. In other words,
the effects of positive circumstantial changes (such as securing a
raise, buying a new car, or moving to a sunnier part of the
country) tend to decay more quickly than the effects of positive
activity changes (such as starting to exercise, changing one’s
perspective, or initiating a new goal or project). This differential
adaptation assumption is rooted in the proposal that circum-
stances (e.g., salary, car ownership, place of residence) represent
relatively static and constant ‘‘facts’’ about one’s life. Thus, al-
though changes in circumstances can trigger increases in well-
being, such boosts tend to be short-lived, because people
quickly begin to take those new circumstances for granted and
cease to derive positive experiences from them. Indeed, hedonic
adaptation to stable situations may explain why life circum-
stances such as income, health status, and geographic region
have been more weakly associated with cross-sectional well-
being than expected (Diener, 1984; Diener et al., 1999) in
cross-sectional studies, participants are undoubtedly sampled at
various lengths of time following the onset of particular circum-
stances, ‘‘watering down’’ the effects of circumstances, overall.
In contrast, intentional activity focuses a person’s energy and
behavior in a variety of different ways, leading to a more
diverse and varied set of experiences, relative to the experiences
produced by circumstances. Also, intentional activity can bring
about an expanding array of new opportunities and possibili-
ties, potentially leading to sustained positive effects in the
person’s life (Fredrickson and Joiner, 2002; Sheldon and
Houser-Marko, 2001). Finally, intentional activity can directly
counteract the tendency toward adaptation, as people might
make the effort to keep varying how and when they engage in
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
60
the activity. These three features of intentional activity should
help to forestall adaptation.
In sum, Lyubomirsky and colleagues’ (2005) model of longi-
tudinal well-being makes clear predictions regarding what
should be found in a three-wave investigation of well-being. Po-
sitive circumstantial changes occurring between Time 1 and
Time 2 should produce enhanced well-being at Time 2, but this
effect should disappear by Time 3, because people quickly
adapt to circumstantial changes, and cease to derive positive
experiences from them. In contrast, positive activity changes be-
tween Time 1 and Time 2 should produce enhanced well-being
at both Time 2 and Time 3. This is because changed activities
are more likely to continue providing positive experiences over
the long-term. These predictions are tested in Studies 2 and 3.
STUDY 1
In Study 1, we set the stage for the longitudinal test by evaluat-
ing an important prediction of Lyubomirsky and colleaguesÕ
(2005) model namely, that hedonic adaptation occurs more
rapidly for circumstantial than for activity change. To this end,
we asked participants to self-select into one of two groups
those who had recently experienced a positive circumstantial
change in their lives, and those who had recently experienced a
positive activity change. All participants rated various charac-
teristics of the change they had experienced, allowing us to
examine between-subject mean differences between the two
groups.
In particular, participants were instructed to rate the extent
to which they have ‘‘gotten used to’’ their change, and, the ex-
tent to which their change ‘‘continues to add variety’’ to their
lives. We expected that activity participants would report hav-
ing adapted less to their change, and would also report deriving
more varied experiences from their change. We also asked par-
ticipants to rate the amount of effort involved in making the
change, and the extent to which the change was intentional ver-
sus accidental. We predicted that activity change participants
would ascribe a greater degree of effort and intentionality to the
SUSTAINABLE HAPPINESS 61
change. Finally, we also asked participants to rate their current
levels of positive affect. We reasoned that the activity change
participants should be happier (assuming that their change
occurred about the same time as that for the circumstantial
change participants), because the well-being associated with
activity change should not have decayed to the same extent as
the well-being associated with circumstantial change.
METHODS
Participants and Procedure
Participants were 224 introductory psychology students, 72
men and 152 women. Two different descriptions of the
experiment were posted on a web-based sign-up system. One
version read: ‘‘Please sign up only if there has been some sig-
nificant positive change in the circumstances of your life since
the beginning of the semester, which has given you a boost
since it occurred. ‘Circumstances’ means ‘facts’ about your
life, such as living arrangement, monetary situation, or
course-load. For example, you may have moved to a better
dorm or better roommate, received an increase in financial
support so you can have more fun, or dropped a course that
you were really going to have trouble with.’’ The other
description read: ‘‘Please sign up only if you have adopted
some significant positive new goal or activity since the begin-
ning of the semester, which has given you a boost since it
occurred. ‘Goal/activity’ means something you chose to do or
get involved in, which takes effort on your part. For exam-
ple, you may have joined a rewarding new group, club, or
sports team, decided on a major or career direction which
makes it clear what to focus on, or taken on some other
important new project or goal in your life.’’
In all, 114 students registered for the activity-change condi-
tion, and 110 registered for the circumstance-change condition.
No participants registered for both sessions. Participants
attended group sessions, which included both conditions.
Participants were given a questionnaire to complete upon
arriving at the testing room. They were first asked to write a
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
62
paragraph about the most important change that had occurred
for them since the beginning of the semester, where the change
had to be either a circumstantial or an activity change, as
defined by the sign-up criteria. Seventy-eight participants were
later excluded (28 in the activity condition and 50 in the cir-
cumstantial condition) because, they did not write about the
condition into which they had previously selected themselves.
We believed this was justified, as we wanted to include only
participants who consistently indicated having experienced ei-
ther one type of change or the other, and exclude participants
who may not have read the sign-up criteria or the session
instructions carefully. Thus, the final sample consisted of 82
activity participants and 64 circumstances participants.
1
Not surprisingly, participants wrote about a wide variety of
changes. Examples of circumstantial changes included ‘‘I
learned that I won’t have to be in a lottery in order to get in
my Broadcast 1 class,’’ ‘‘My roommate at the beginning of the
semester was a cocaine addict; she is no longer my roommate,’’
‘‘This week I found out that I received a scholarship that I
wasn’t expecting at all,’’ ‘‘My mother was recently diagnosed
with brain cancer and a couple of days ago a surgery was per-
formed and everything turned out fine,’’ and ‘‘I was recently
initiated into my fraternity...I no longer have to worry about
initiation requirements.’’ Examples of activity changes included
‘‘When I first got here my classes seemed hard and I didn’t
study as much as I should have. I set myself a goal to study for
at least 5 h a day and now my classes are going a lot better for
me,’’ ‘‘I enrolled in a class that is helping me to figure out a
correct career choice for me,’’ ‘‘I used to not ever go to church,
but now I am going to Campus Crusade for Christ meetings,
and God is more a part of my life than He ever has been,’’ ‘‘I
have started lifting weights 4 or 5 days a week. This has drasti-
cally boosted my energy levels,’’ and ‘‘I made a goal for myself
that I would get involved and spend mostly all of my free time
working on homecoming for my fraternity.’’
Measures
After describing the life change, participants made a number of
ratings of the particular change they had experienced. First, we
SUSTAINABLE HAPPINESS 63
assessed the extent to which the change had been intentional,
and the extent to which the change required effort to bring
about. We asked these questions in order to help validate the
manipulations, since changes based in volitional activity should
be perceived as more intentional and more effortful compared
to changes rooted in external facts and circumstances. Specifi-
cally, to assess the intentionality of the change, we combined
the following two questions: ‘‘To what extent was the change
your own doing? That is, how much did you bring it about?’’
and ‘‘To what extent did you choose to have this change? That
is, how much did you mean for it to happen?’’ (a=0.74). To
assess the degree of effort associated with the change, we com-
bined the following two questions: ‘‘How much effort did you
have to put into making the change happen?’’ and ‘‘How
much effort do you have to put into keeping the change
going?’’ (a=0.61). A 1 (not at all)to5(very much) scale was
used for each rating.
To assess hedonic adaptation, we used the following question:
‘‘To what extent have you gotten used to the change? That is, to
what extent do you find that you’ve become accustomed to it,
such that it doesn’t give the same boost as before?’’ To assess
experiential variety, we employed the following question: ‘‘To
what extent is the change something that varies over time i.e.,
something that adds variety to your life?’’ These two variables
allowed us to test our primary theoretical hypotheses, relevant
to understanding the sustainable effects of activity and circum-
stantial change upon well-being.
To assess positive mood, we administered the positive affect
scale of the PANAS (Watson et al., 1988). This 10-item scale
presents a variety of positive mood adjectives, such as ‘‘in-
spired’’, ‘‘alert,’’ and ‘‘strong.’’ Participants were asked to rate
how the change they wrote about had affected their typical feel-
ings since it occurred, using a 1 (much less of this feeling)to3
(same)to5(much more of this feeling) scale.
Finally, participants rated how long ago the change had
occurred, using a scale ranging from 1 (110 days ago)to4(30
or more days ago). This question was included to evaluate one
alternative explanation of any activity versus change effects
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
64
namely, that the two types of reported change differed in how
long ago they had occurred.
RESULTS
We conducted a series of independent sample t-tests to com-
pare the means of the two groups (see Table I). The analyses
revealed significant group differences for the intentionality vari-
able, t(144) = 4.31, p<0.01, such that the activity group judged
the change in their lives as relatively more of a product of their
own intentions. A significant mean difference was also found
for the effort variable, t(144) = 3.91 p<0.01, such that the ac-
tivity group mean was higher. Again, these results may be
viewed as validating our measures of activity-based versus cir-
cumstance-based life change as, according to our assumptions,
changes rooted in volitional activity should be perceived as
more intentional and more effortful compared to changes
dependent on external facts and circumstances.
We then turned to our theoretically derived hypotheses. As
expected, the activity group reported acquiring much more var-
ied experiences from their change, t(144) = 2.46, p<0.05,
whereas the circumstances group reported a relatively stronger
sense of ‘‘having gotten used to’’ their change, t(144) = 3.67,
p<0.01. Furthermore, relative to the circumstance group, the
activity group reported that the positive change they had
TABLE I
Mean differences between the activity change group and the circumstantial
change group (Study 1)
Activity change Circumstantial change
Validity checks
Intentionality 4.30 3.68
Effort 3.95 3.31
Hypothesis tests
Habituation 2.96 3.57
Variety 3.43 2.90
Positive affect 4.03 3.67
Note: All group differences are significant at the 0.01 level or higher.
SUSTAINABLE HAPPINESS 65
experienced had produced a greater increase in positive affect,
t(144) = 4.25, p< 0.01. Given that the scale referred to change
in affect with ‘‘3’’ representing no change, and also the fact that
both means were greater than 3.5, it appears that both groups
indeed experienced a ‘‘boost’’ from their respective change.
However, the boost was not as large, or as long-lasting, for the
circumstance change group (see Table I).
Might the group differences reported above have resulted be-
cause activity changes had generally occurred more recently
than circumstantial changes? No, as the two groups did not dif-
fer on this variable, t(144)=1.43, ns. The sample mean for the
‘‘time elapsed’’ variable was 2.92, indicating that participants’
changes occurred approximately 30 days ago, on average.
Finally, although some significant main effects of gender
emerged in some of the analyses reported above, gender did not
interact with change-type, suggesting that the processes specified
by our model are the same for both sexes.
BRIEF DISCUSSION
Study 1 provided preliminary evidence for our dynamic process
model, by supporting a central assumption of that model that
circumstantial changes are characterized by more hedonic adap-
tation, and, conversely, that activity changes give rise to more
experiential variety. In addition, Study 1 confirmed that activity
changes are more intentional and effortful. Finally, this study
revealed that activity change participants experienced greater
positive affect as a result of their change, compared to the cir-
cumstance change participants. Given the fact that intentional
activities are more volitional in nature and less susceptible to
hedonic adaptation, it appears that such activities may indeed
offer a promising route to ‘‘the pursuit of happiness.’’
One limitation of Study 1 is that it was cross-sectional, not
longitudinal. Thus, Study 1 did not address change in well-
being, the central outcome in our model. Also, this study could
not address the possibility that stable personality variables ac-
count for the patterns observed that is, whether participants’
willingness to self-select into one group over another reflects
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
66
some individual difference that also happened to be associated
with the concurrent outcome measures. To remedy these limita-
tions, we conducted a longitudinal study, which gave us an
opportunity to assess concrete change in well-being, as well as
to partial out stable personality differences that should have
equal influence at every measurement period.
STUDY 2
Study 2 was designed to provide a direct test of our dynamic
process model, by comparing the longitudinal effects of circum-
stantial change and activity change on gains in well-being in a
three-wave study of a large sample of individuals. Well-being
was measured at all three waves, as were positive activity chan-
ges and positive circumstantial changes occurring between
Times 1 and 2. Thus, unlike the previous study, this study as-
sessed the magnitude of activity and circumstantial changes
within the same participants, allowing us to use a simultaneous
regression strategy. We hypothesized that both types of change
would have effects on well-being at Time 2, but that only activ-
ity change would still have effects at Time 3. We also examined
mean-level changes in well-being (relative to the participants’
own past levels), in addition to rank-order shifts in well-being
(relative to others in the sample), as these two types of change
are orthogonal and perhaps equally important means of evalu-
ating temporal changes in well-being.
METHOD
Participants and Procedure
Participants were 669 students in psychology classes at a large
state university, 249 men and 420 women, who took part in ex-
change for extra credit. The data were collected during three
different semesters, from three separate samples, using the same
study design in each case. Because some mean differences
emerged among samples, we control for sample membership
using two dummy variables in our primary hypothesis tests,
below.
SUSTAINABLE HAPPINESS 67
Students completed three in-class questionnaires, one near
the beginning of the semester (Time 1), one near the mid-point
of the semester (6 weeks later; Time 2), and one near the end of
the semester (12 weeks later; Time 3). Participants rated recent
activity and circumstantial changes within their lives at Time 2,
and rated their current well-being at Times 1, 2, and 3.
Measures
To extend our assessment of well-being, we employed the full
20-item PANAS (Watson et al., 1988). This enabled us to exam-
ine negative affect (NA; i.e., ‘‘irritable,’’ ‘‘ashamed’’), as well as
positive affect (PA). Alpha coefficients for negative affect ranged
between 0.79 and 0.87 over three assessments, and alpha coeffi-
cients for positive affect ranged between 0.82 and 0.85. We also
administered the 5-item Satisfaction With Life Scale (SWLS;
Diener et al., 1985), to assess this more cognitive aspect of well-
being (e.g., ‘‘The conditions of my life are excellent’’; alphas
ranged between 0.81 and 0.86). Together, these three variables
PA, NA, and life satisfaction form the core of subjective
well-being, according to Diener (1984, 1994). At each assess-
ment, participants rated their well-being ‘‘right now in your
life,’’ using 1 (not at all)to5(very much) scales. To provide the
most reliable measure of well-being, we also created an aggre-
gate SWB variable at each time period, by summing PA and life
satisfaction and subtracting NA. This procedure has been used
successfully in many past studies (e.g., Diener, 1994; Elliot and
Sheldon, 1996; Sheldon and Elliot, 1999; Sheldon and Kasser,
1995). Supporting its use, factor analyses of PA, life satisfac-
tion, and NA (reversed) revealed a single factor at each time
period, which accounted for 54%, 57%, and 55% of the vari-
ance, respectively.
In order to assess activity change and circumstantial change,
we used the same two paragraphs that were employed to re-
cruit the Study 1 participants. Specifically, at Time 2, the par-
ticipants in Study 2 rated the extent to which they had
experienced both an activity change and a circumstantial
change since Time 1, as defined by those two paragraphs
(1 = no positive change;5=much positive change). Again, we
intended to use these ratings in simultaneous regressions, so
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
68
that we could compare the relative impact of activity change
and circumstantial change.
RESULTS
Preliminary Analyses
Table II presents descriptive statistics for the primary study
variables. Participants reported higher levels of positive activity
change than positive circumstantial change, t(668) = 9.36,
p<0.01. Also, there was a significant sample-wide decline in ag-
gregate SWB from Time 1 to Time 2, t(668) = 5.45, p<0.01, a
decline that was also reflected in the comparison between Time
1 and Time 3, t(668) = 4.32, p<0.01. These effects were driven
by significant decreases in PA and significant increases in NA
between Time 1 and Times 2 and 3. As a final preliminary ana-
lysis, we correlated the activity change and circumstantial
change variables, finding an association of 0.31 (p<0.01). This
association indicates that there is some overlap between the two
constructs, but not a large amount; thus, it is reasonable to
examine their simultaneous influence.
TABLE II
Descriptive statistics for Study 2 measures
Mean Standard deviation
Positive activity change 3.39 1.08
Positive circumstantial change 2.92 1.10
T1 Aggregate well-being 4.93 1.66
T2 Aggregate well-being 4.67 1.67
T3 Aggregate well-being 4.69 1.73
T1 Positive affect 3.59 0.66
T2 Positive affect 3.55 0.65
T3 Positive affect 3.49 0.71
T1 Negative affect 2.09 0.76
T2 Negative affect 2.28 0.76
T3 Negative affect 2.27 0.77
T1 Life satisfaction 3.44 0.82
T2 Life satisfaction 3.40 0.79
T3 Life satisfaction 3.47 0.83
SUSTAINABLE HAPPINESS 69
Rank-Order Shifts
Next, we tested our hypothesis that both positive activity
change and positive circumstantial change would predict chan-
ges in well-being from Time 1 to Time 2. Specifically, we
regressed each Time 2 well-being variable (SWB, PA, NA, and
life satisfaction) upon the corresponding Time 1 well-being vari-
able and also the two change variables. Table III provides the
standardized coefficients that resulted. As expected, Time 1 well-
being was highly significant in every analysis (all p’s<0.01),
indicating considerable rank-order stability, presumably reflect-
ing the set point and stable personality differences. Activity
change was also significant in every analysis, and circumstantial
change was significant in all but the NA analysis.
Next, we tested our primary hypothesis that is, that posi-
tive activity change would predict well-being at Time 3, whereas
activity change would not. Specifically, we regressed the Time 3
well-being variables on the corresponding Time 1 well-being
variables and also the two change variables. Table III also
TABLE III
Standardized coefficients predicting change in well-being (Study 2)
Time 2 Time 3
Subjective well-being
Time 1 subjective well-being 0.66* 0.56*
Being
Activity change 0.12* 0.12*
Circumstantial change 0.10* 0.01
Positive affect
Time 1 positive affect 0.48* 0.44*
Activity change 0.16* 0.16*
Circumstantial change 0.16* 0.05
Negative affect
Time 1 negative affect 0.54* 0.45*
Activity change )0.09* )0.05
Circumstantial change )0.05 0.03
Life satisfaction
Time 1 life satisfaction 0.74* 0.68*
Activity change 0.07* 0.10*
Circumstantial change 0.07* 0.03
*p< 0.01.
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
70
contains these results. Time 1 well-being was again highly sig-
nificant in every analysis (all p’s<0.01). However, as predicted,
only activity change was significant in the SWB analysis,
whereas circumstantial change fell to non-significance. Circum-
stantial change was also non-significant for each individual
SWB variable, whereas activity change was significant for PA
and life-satisfaction but not NA.
Mean-Level Shifts
The analyses reported above concerned rank-order shifts in
well-being. Next, we considered the within-subject means that
resulted from the two types of change, focusing on the aggre-
gate SWB variable. First, we created two difference scores by
subtracting Time 1 SWB from Time 2 SWB, and Time 1 SWB
from Time 3 SWB. This allowed us to consider each partici-
pant’s variations from his or her initial SWB baseline at each of
the two follow-up periods. Next, we normalized the activity
change and circumstantial change variables from a 5-point to a
3-point scale, thereby creating balanced groups in the low and
high ends of each distribution.
2
We then conducted two ANOVAs, one for each change vari-
able, in which amount of activity or circumstantial change (low,
medium, or high) was a between-subjects factor with three lev-
els, and comparison period (from T1 to T2, or from T1 to T3)
was a within-subjects factor with two levels. The two panels of
Figure 1 present the means produced in each analysis. As can
be seen, in the activity change analysis (top panel), the low and
medium change groups declined in SWB between Time 1 and
Time 2, and stayed at this lower level at Time 3. However, the
high activity change group essentially maintained their initial
levels, at both Time 2 and Time 3. There were no significant
main effects of amount of change or of time period, and no sig-
nificant interaction between the two factors (all p’s>0.10). In
the circumstantial change analysis (bottom panel), the low
change group declined the most in SWB from Time 1 to Time
2, then somewhat bounced back. The medium change group
declined somewhat less, then maintained the lower level. The
high change group declined the least at first, but then declined
more from Time 2 to Time 3. These variable patterns of change
SUSTAINABLE HAPPINESS 71
Figure 1. Longitudinal differences in SWB by amount of life change and
type of life change positive activity change (top panel) and positive cir-
cumstance change (bottom panel) (Study 2).
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
72
were reflected in a significant Amount of Change Time Peri-
od interaction, F
(2,666)
=5.28, p<0.01; neither main effect was
significant.
BRIEF DISCUSSION
Study 2 provided support for our dynamic process model of
longitudinal well-being. Although positive circumstantial change
predicted a positive rank-order shift in well-being at Time 2, it
was not associated with well-being at a third assessment, near
the end of the semester. Again, we assume that this occurred in
part because of hedonic adaptation to positive circumstantial
change, such that participants no longer noticed or benefited
from these changes. In contrast, activity change predicted rank-
order gains between Time 1 and Time 2 and also predicted
maintenance of these gains at Time 3. Again, we assume that
this occurred in part because of the steady stream of positive
experiences induced by activity change. Notably, the two types
of life change exerted larger effects on enhancing positive well-
being (i.e., by promoting rank-order increases in positive mood
and life satisfaction), than they did on ameliorating negative
well-being (i.e., by promoting rank-order decreases in negative
mood).
Examination of mean-level patterns revealed that the highest
positive activity change group did not actually succeed in boost-
ing their levels of well-being instead, they largely maintained
their initial levels. Although sustaining one’s previous level of
well-being may not seem as desirable or noteworthy as increas-
ing one’s well-being beyond one’s previous level, we believe that
the ability to ‘‘maintain the initial glow’’ of the semester is a sig-
nificant achievement. In other words, amid circumstances in
which one’s cohort is experiencing stress and declining well-
being, our model may best describe a process of resilience by
which this general pattern is avoided. In contrast, when one’s
cohort is experiencing ‘‘good times’’ and sample-wide increases
in well-being (e.g., during summer vacation), presumably, our
model would pinpoint those who experienced the largest boosts
of all.
SUSTAINABLE HAPPINESS 73
STUDY 3
In Study 3, we sought to replicate the basic Study 2 findings
concerning changes in SWB (Diener, 1984, 1994). However, we
included additional measures of well-being, to evaluate the
breadth of the phenomenon identified in Study 2. Specifically,
we additionally assessed psychological well-being (PWB; Ryff
and Keyes, 1995; Keyes et al., 2002), which includes more glo-
bal measures of personal and psychosocial thriving, including
relationship-satisfaction, mastery, purpose, self-acceptance, per-
sonal growth, and autonomy. Keyes and his colleagues (2002)
have shown that these constructs perform somewhat differently
than the conventional SWB measures employed in Study 2.
However, we believed that our model should apply to PWB as
well, because activity changes can presumably enhance one’s
sense of mastery, relatedness, and purpose. Finally, we included
Lyubomirsky and Lepper’s (1999) measure of subjective happi-
ness, to directly address the question posed by the title of this
article and to tap into a more global, molar well-being con-
struct, which more closely matches what most laypeople are try-
ing to pursue. We expected our model to apply to this measure
as well.
As a further change in Study 3, we broadened our assess-
ment of activity and circumstantial changes beyond the single
item measures used in Study 2. That is, changes in specific
types of circumstances and specific types of activities were as-
sessed. In so doing, we hoped to further validate our
assumption that these represent two distinct categories of life
change.
METHOD
Participants and Procedure
Participants were 275 students in a psychology class, who took
part for extra course credit. As in Study 2, questionnaires were
administered in class at three points during the semester. The
final sample for the study comprised 204 participants (136
women and 68 men), who completed all three questionnaires.
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
74
Measures
As in Study 2, participants completed the PANAS and the
SWLS at three points during the semester, and an aggregate
SWB variable was computed for each time of measurement by
adding PA and life satisfaction, and subtracting NA. Again,
factor analyses supported this procedure, as a single factor was
found at each time period, accounting for 64%, 59%, and 59%
of the variance, respectively. In addition, participants completed
the 4-item Subjective Happiness Scale at each time period
(Lyubomirsky and Lepper, 1999; a’s =0.80, 0.80, and 0.81).
Finally, on each occasion, participants completed the PWB
scale (Ryff and Keyes, 1995; Keyes et al., 2002), an 18-item
measures of psychosocial thriving (a’s = 0.78, 0.81, and 0.83).
To assess both circumstantial and activity change between
Time 1 and Time 2, we administered the same paragraphs em-
ployed in Study 2. However, for each change variable, we also
included three other, more specific, items. For circumstantial
change, these new items were as follows: ‘‘To what extent has
your monetary situation changed in the past 6 weeks (i.e., you
have gained or lost money)?,’’ ‘‘To what extent have your living
arrangements changed in the past 6 weeks (i.e., you live in a
better/worse place, or with a better/worse roommate)?,’’ and
‘‘To what extent has your relationship status changed in the
past 6 weeks (i.e., you have acquired or lost a boyfriend/
girlfriend)?’’ For activity change, the additional three items were
the following: ‘‘To what extent have your goals, projects, or
strivings changed in the past 6 weeks?’’; ‘‘To what extent has
your diet, exercise, or other self-maintenance activity changed in
the past 6 weeks?’’; and ‘‘To what extent have your conscious
attitudes or mental approach to life changed in the past
6 weeks?’’ For each change variable, four items (the additional
three items, plus the item used in Study 2) were averaged to
form an aggregate circumstantial change variable and an aggre-
gate activity change variable.
A principal components analysis of the eight items revealed a
clean two-factor solution, with each item loading on the ex-
pected factor at 0.56 or more, with no cross-loading greater
than 0.24. These analyses support our contention that activity
and circumstantial changes are empirically distinct constructs.
SUSTAINABLE HAPPINESS 75
Still, the two factors only accounted for 44% of the variance,
and the alpha coefficients were low for the two composites
(a’s = 0.60 and 0.47, respectively). However we believe this is
understandable given that people would not be expected to
report experiencing every activity change or every circumstantial
change on the list. What is most important is the number of
such changes endorsed, and the fact that the two types of
change were empirically distinct. Nunnally and Bernstein (1994)
coined the term ‘‘emergent construct’’ to refer to such measures,
which additively combine items that may not be highly corre-
lated themselves.
RESULTS
Preliminary Analyses
As in Study 2, we first assessed sample-wide well-being changes
over the semester. Table IV contains the relevant means. In
contrast to Study 2, no sample-wide changes were observed (all
t’s<1.52). In a second preliminary analysis, the activity change
and circumstantial change variables were found to have a weak-
er association than in Study 2 (r= 0.20, p<0.01), further sug-
gesting that they are distinct constructs, especially when
measured with multiple items.
TABLE IV
Descriptive statistics for Study 3 measures
Mean Standard deviation
Positive activity change 3.60 0.56
Positive circumstantial change 3.10 0.59
T1 Subjective well-being 5.10 1.86
T2 Subjective well-being 5.14 1.64
T3 Subjective well-being 4.98 1.84
T1 Happiness 3.57 0.77
T2 Happiness 3.59 0.74
T3 Happiness 3.63 0.77
T1 Psychological well-being 3.99 0.42
T2 Psychological well-being 3.99 0.42
T3 Psychological well-being 3.98 0.45
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
76
Primary Analyses
Rank-order analyses
We tested the same regression models as in Study 2, separately
for SWB, PWB, and happiness. As shown in Table V, both activ-
ity changes and circumstantial changes predicted rank-order
gains in SWB at Time 2, controlling for Time 1. In addition, both
types of changes predicted rank-order gains in happiness at Time
2. However, only activity changes reached significance in predict-
ing changes in PWB at Time 2, whereas circumstantial changes
did not predict enhanced PWB. Turning to Time 3, only activity
changes predicted maintained gains at Time 3, for all three out-
comes; circumstantial changes were non-significant in every case.
Thus, these data replicate the dynamic patterns shown in Study
2, extending them to two new measures of well-being.
Mean-level analyses
Next, we examined the issue of mean-level change in SWB, hap-
piness, and PWB. However, because activity and circumstantial
changes were measured via multi-item composites instead of
single items, we could not employ the categorical ANOVA ap-
proach described in Study 2. Instead, we used the regression
TABLE V
Standardized coefficients predicting change in well-being
(Study 3)
Time 2 Time 3
Subjective well-being
Time 1 Subjective well-being 0.64* 0.48*
Activity change 0.19* 0.19*
Circumstantial change 0.12* 0.03
Happiness
Time 1 happiness 0.64* 0.59*
Activity change 0.24* 0.18*
Circumstantial change 0.16* 0.07
Psychological well-being
Time 1 psychological well-being 0.71* 0.68*
Activity change 0.15* 0.15*
Circumstantial change 0.06 0.06
*p60.01.
SUSTAINABLE HAPPINESS 77
models described above to compute predicted means for two
representative persons: one who started out at Time 1 with a
level of SWB equal to the sample mean who then experienced
activity changes one standard deviation above the mean and
circumstantial changes at the mean, and another who also star-
ted out at the sample mean who then experienced circumstantial
changes one standard deviation above the mean and activity
changes at the mean. This enabled us to examine the trajecto-
ries of those who made significant activity changes in combina-
tion with an average degree of circumstantial change, and vice
versa.
Figure 2 contains the SWB trajectories for those making
above-average circumstantial changes and average activity
changes, and for those making above-average activity changes
and average circumstantial changes. As can be seen, those
making above-average activity changes increased in their mean
levels of SWB, and then maintained most of the gain. Those
making above-average circumstantial changes increased in their
mean levels of SWB, then lost all of that gain. The pattern for
the happiness construct was quite similar to that depicted in
Figure 2, with a peak and subsequent decline for those making
Figure 2. Predicted changes in SWB for those reporting very positive activ-
ity changes at Time 2, and those reporting very positive circumstantial chan-
ges at Time 2 (Study 3).
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
78
circumstantial changes, and a maintained peak for those mak-
ing activity changes. The pattern for the PWB construct was
also the same for the activities group, but was somewhat dif-
ferent for the positive circumstantial changes group, which
manifested a flat trajectory throughout the semester on this
measure.
BRIEF DISCUSSION
Study 3 replicated Study 2’s most important finding namely,
that activity changes lead to maintainable rank-order increases
in SWB, and circumstantial changes do not. In addition, Study
3 generalized the basic effect to measures of subjective happi-
ness (Lyubomirsky and Lepper, 1999) and psychological or psy-
chosocial well-being (Ryff and Keyes, 1995). Furthermore,
Study 3 broadened the measurement of the activity change and
circumstantial change constructs. Finally, Study 3 suggested
that activity changes can lead to sustainable gains in absolute
levels of well-being, as well as sustainable rank-order gains in
well-being.
GENERAL DISCUSSION
These studies tested an important proposition of Lyubomirsky
and colleagues’ (2005) longitudinal model of well-being. Again,
the model distinguishes between two types of beneficial life
changes, circumstantial changes and activity changes, reflecting
theoretical distinctions often employed in the SWB literature.
Although both types of change are assumed by the model to
positively influence well-being in the short-term, the effects of
circumstantial change are postulated to be less enduring,
because of hedonic adaptation. The current studies found clear
and consistent support for this prediction, not only for sub-
jective well-being (positive affect, negative affect, and life satis-
faction), but also, for subjective happiness (Lyubomirsky and
Lepper, 1999), and for psychological well-being (Ryff and Keyes,
1995).
SUSTAINABLE HAPPINESS 79
We believe these findings have important implications for a
controversial, and under-studied, issue in well-being research
(Lucas et al., 2003). Can a person sustainably increase her level
of well-being, or is she doomed to return to her genetic set
point, after any temporary change? If the latter perspective is
correct, then positive psychology research, the ‘‘self-help’’ move-
ment, and the American dream more generally may offer only
empty promises. However, if well-being boosts are indeed sus-
tainable over the long-term, then it becomes imperative to
understand how this may be done. The current studies offer
promising new information concerning this question.
Can positive circumstantial changes have no sustainable im-
pact on well-being? This conclusion is probably an overstate-
ment. First, it is undoubtedly necessary to achieve a threshold
of basic positive circumstances before one can concentrate on
the ‘‘better things’’ in life (Oishi et al., 1999). In other words,
under extremely unfavorable or impoverished conditions,
changing one’s circumstances (i.e., acquiring a safe place to live
and a secure source of sustenance) can have large effects on
well-being, although such changes may only get one to one’s set
point, and not above it. Second, even assuming that an individ-
ual has already met his basic needs, it may be possible for him
to further improve his well-being via further circumstantial
changes. However, we believe this is feasible only to the extent
that one takes action to keep the new circumstances ‘‘fresh’’
i.e., by remembering to appreciate or feel gratitude for them, or
by making the effort to take advantage of the opportunities for
positive experiences that they afford. In other words, this is fea-
sible when one engages in intentional activity with respect to
the circumstances in one’s life that is, when one acts upon
one’s circumstances.
How sharp is the distinction between activity change and cir-
cumstantial change? This is an important question, because
obviously, many circumstantial changes are brought about by
intentional activity, and many intentional activities are depen-
dent on circumstances. Despite the potentially fuzzy boundary
between them, the two constructs were not highly correlated
(r’s = 0.31 and 0.20 in Studies 2 and 3), and also yielded dis-
tinct and differential results in the current research, as predicted
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
80
by our theory and past findings in the well-being literature. In-
deed, the clear and strong group differences found in Study 1
suggest that our participants had little difficulty understanding
and distinguishing between circumstances and activities.
Perhaps what is most critical is not the objective life change
that occurs, but, rather, each individual’s subjective construal of
that change. That is, the same event (e.g., getting married) can
signify to one person a circumstance to which she adapts quick-
ly, but, to another person, an intentional activity that can give
rise to a diverse stream of positive experiences.
Limitations and Remaining Questions
These studies have several limitations. First, we only examined
three waves of data in Studies 2 and 3, and the full model test
focused on a relatively short period of time (i.e., one semester).
Thus, it remains to be seen whether the positive activity effects
demonstrated herein would persist across further waves of data,
collected over a longer period of time. A second limitation was
that we did not evaluate the many possible moderators of the
activity-change effects. Adopting a new goal or program of
activity may not do one any good if the activity does not fit
one’s needs, predispositions or talents, if one pursues the activ-
ity for the ‘‘wrong’’ reasons, if one fails to do well at the activ-
ity, or if one practices the activity in a monothematic or
invariant way. In short, it is undoubtedly necessary to select
appropriate activities, to attend to the optimal timing, variety,
and other factors in the way that one practices them, and to do
reasonably well at them (Lyubomirsky et al., in press). We be-
lieve such unmeasured moderators can help explain the some-
what modest change main effects observed herein, and represent
a promising future research direction.
Third, the present research does not establish the generaliz-
ability of the effects to people of other ages, socioeconomic
statuses, or cultural memberships. Perhaps such factors affect
the ease with which people are able to make activity-based life
changes, or affect how much influence circumstantial and
activity-based changes have on peoples’ well-being. Further-
more, future researchers should collect more objective mea-
sures of well-being, involving peer report or objective
SUSTAINABLE HAPPINESS 81
psychosocial functioning. Finally, it would be fruitful in the
future to expand the assessment of activity and circumstantial
changes, to better represent these important categories of
effects upon well-being.
Finally, we did not attempt to measure the happiness set
point in this research, although it is an important part of our
theoretical model. We assumed that the set point’s influence is
largely constant and is represented by each participant’s mean
level across the multiple well-being assessments. However, fu-
ture researchers might attempt to explicitly model the set point,
perhaps using some combination of existing personality mea-
sures, or by measuring an individual’s average well-being over
many years (Lykken, 1999), or perhaps by assessing a weighted
average of the well-being of a given participant’s biological fam-
ily members. Although using such measures to predict mean
levels of well-being over time is a potentially important enter-
prise, it is one that is largely orthogonal to the current research,
which focused on variations in well-being over time.
CONCLUSION
A central presumption of the ‘‘American dream’’ is that,
through their own efforts and hard work, people may move to-
wards greater happiness and fulfillment in life. This assumption
is echoed in the writings of philosophers, both ancient and
modern. In Nicomachean Ethics, Aristotle (1985) proposed that
happiness involves engagement in activities that promote one’s
highest potentials. And, in the Conquest of Happiness, Bertrand
Russell (1930/1975) argued that the secrets to happiness include
enterprise, exploration of one’s interests, and the overcoming of
obstacles. The current studies provide important new support
for these ideas. But do they also glorify the ideals of pleasure
and self-gratification, which are also endemic to American
culture? We believe not. Indeed, the fact that activity changes
require considerable effort to enact is more consistent with the
Puritan version of the American dream, rather than the ‘‘easy
living’’ or the ‘‘quick fix’’ ideals that have partially supplanted
this foundational vision. In other words, our data suggest
that effort and hard work offer the most promising route to
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
82
happiness. In contrast, simply altering one’s superficial circum-
stances (assuming they are already reasonably good) may have
little lasting effect on personal well-being. We believe these are
potentially potent prescriptions to keep in mind.
NOTES
1
Notably, the effects reported below are still significant, albeit somewhat
weaker, when the full sample is used.
2
Specifically, our change variables were recoded as follows: activity change
variable (1=1; 2=1; 3=2; 4=2; 5=3) and circumstantial change variable
(1=1; 2=1; 3=2; 4=3; 5=3). This resulted in ns of 116, 447, and 106 for
three levels of the activity change variable, and ns of 216, 269, and 184 for
three levels of the circumstantial change variable.
REFERENCES
Aristotle, M.: 1985, Nicomachean Ethics. (T. Irwin, Trans.) (Hackett, India-
napolis IN).
Argyle, M.: 1999, Causes and correlates of happiness, in D. Kahneman
E. Diener and N. Schwarz (eds), Well-being: The Foundations of Hedonic
Psychology (Russell Sage Foundation, New York), pp. 353373.
Atkinson, T.: 1982, ÔThe stability and validity of quality of life measuresÕ,
Social Indicators Research 10, pp. 113132.
Bandura, A.: 1997, Self-efficacy: The Exercise of Control (Freeman, New
York).
Brickman, P. and D.T. Campbell: 1971, Hedonic relativism and planning the
good society, in M.H. Appley (ed.), Adaptation-level Theory (Academic
Press, New York), pp. 287302.
Brickman, P., D. Coates and R. Janoff-Bulman: 1978, ÔLottery winners and
accident victims: Is happiness relative?Õ, Journal of Personality and Social
Psychology 36, pp. 917927.
Brunstein, J.C., O.C. Schultheiss and R. Grassman: 1998, ÔPersonal goals and
emotional well-being: The moderating role of motive dispositionsÕ, Journal
of Personality and Social Psychology 75, pp. 494508.
DeNeve, K.M. and H. Cooper: 1998, ÔThe happy personality: A meta-analysis
of 137 personality traits and subjective well-beingÕ, Psychological Bulletin
124, pp. 197229.
Diener, E.: 1984, ÔSubjective well-beingÕ, Psychological Bulletin 95, pp.
542575.
SUSTAINABLE HAPPINESS 83
Diener, E.: 1994, ÔAssessing subjective well-being: Progress and opportunitiesÕ,
Social Indicators Research 31, pp. 103157.
Diener, E., R. Emmons, R. Larsen and S. Griffin: 1985, ÔThe Satisfaction with
Life ScaleÕ, Journal of Personality Assessment 47, pp. 11051117.
Diener, E. and R.E. Lucas: 1999, Personality and subjective well-being, in
D. Kahneman, E. Diener and N. Schwartz (eds), Well-being: The Founda-
tions of Hedonic Psychology (Russell Sage, New York), pp. 213229.
Diener, E., E.M. Suh, R.E. Lucas and H.L. Smith: 1999, ÔSubjective well-being:
Three decades of progressÕ, Psychological Bulletin 125, pp. 276302.
Diener, E., E.M. Suh, H. Smith and L. Shao: 1995, ÔNational differences in
reported well-being: Why do they occur?Õ, Social Indicators Research 34,
pp. 732.
Dijkers, M.: 1997, ÔQuality of life after spinal cord injury: A meta-analysis of
the effects of disablement componentsÕ, Spinal Cord 32, pp. 829840.
Elliot, A.E. and K.M. Sheldon: 1996, ÔAvoidance achievement motivation: A
personal goals analysisÕ, Journal of Personality and Social Psychology 7,
pp. 667679.
Emmons, R.A. and L.A. King: 1988, ÔConflict among personal strivings:
Immediate and long-term implications for psychological and physical well-
beingÕ, Journal of Personality and Social Psychology 54, pp. 10401048.
Emmons, R.A. and M.E. McCullough: 2003, ÔCounting blessings versus bur-
dens: An experimental investigation of gratitude and subjective well-being in
daily lifeÕ, Journal of Personality and Social Psychology 84, pp. 377389.
Fava, G.: 1999, ÔWell-being therapy: Conceptual and technical issuesÕ,
Psychotherapy and Psychosomatics 68, pp. 171179.
Fordyce, M.W.: 1977, ÔDevelopment of a program to increase happinessÕ,
Journal of Counseling Psychology 24, pp. 511521.
Fordyce, M.W.: 1983, ÔA program to increase happiness: Further studiesÕ,
Journal of Counseling Psychology 30, pp. 483498.
Frederick, S. and G. Loewenstein: 1999, Hedonic adaptation, in D. Kahn-
eman, E. Diener and N. Schwarz (eds), Well-being: The foundations of
hedonic psychology (Russell Sage Foundation, New York), pp. 302329.
Fredrickson, B.L. and T. Joiner: 2002, ÔPositive emotions trigger upward spi-
rals toward emotional well-beingÕ, Psychological Science 13, pp. 172175.
Freedman, J.: 1978, Happy people: What happiness is, who has it, and why
(Harcourt Brace Jovanovich, New York).
Headey, B. and A. Wearing: 1989, ÔPersonality, life events, and subjective well-
being: Toward a dynamic equilibrium modelÕ, Journal of Personality and
Social Psychology 57, pp. 731739.
Kahneman, D.: 1999, Objective happiness, in D. Kahneman, E. Diener
N. Schwarz (eds), Well-being: The Foundations of Hedonic Psychology
(Russell Sage Foundation, New York), pp. 325.
Kasser, T. and R.M. Ryan: 1993, ÔA dark side of the American dream: Cor-
relates of financial success as a central life aspirationÕ, Journal of Personality
and Social Psychology 65, pp. 410422.
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
84
Kasser, T. and R.M. Ryan: 1996, ÔFurther examining the American dream:
Differential correlates of intrinsic and extrinsic goalsÕ, Personality and Social
Psychology Bulletin 22, pp. 280287.
Keyes, C.L.M., D. Shmotkin and C.D. Ryff: 2002, ÔOptimizing well-being: The
empirical encounter of two traditionsÕ, Journal of Personality and Social
Psychology 82, pp. 10071022.
King, L.A.: 2001, ÔThe health benefits of writing about life goalsÕ, Personality
and Social Psychology Bulletin 27, pp. 798807.
Lichter, S, K. Haye and R. Kammann: 1980, ÔIncreasing happiness through
cognitive retrainingÕ, New Zealand Psychologist 9, pp. 5764.
Lucas, R.E., A.E. Clark, Y. Georgellis and E. Diener: 2003, ÔReexamining
adaptation and the set point model of happiness: Reactions to changes in
marital statusÕ, Journal of Personality and Social Psychology 84, pp.
527539.
Lykken, D.: 2000, Happiness: The nature and nurture of joy and contentment
(St. Martin’s Griffin, New York).
Lykken, D. and A. Tellegen: 1996, ÔHappiness is a stochastic phenomenonÕ,
Psychological Science 7, pp. 186189.
Lyubomirsky, S. and H.S. Lepper: 1999, ÔA measure of subjective happiness:
Preliminary reliability and construct validationÕ, Social Indicators Research
46, pp. 137155.
Lyubomirsky, S. and L. Ross: 1997, ÔHedonic consequences of social com-
parison: A contrast of happy and unhappy peopleÕ, Journal of Personality
and Social Psychology 73, pp. 11411157.
Lyubomirsky, S., Sheldon, K.M., Schkade, D., 2005, ‘Pursuing happiness: The
architecture of sustainable change’, Review of General Psychology 9,
pp. 111131.
Lyubomirsky, S., Sousa, L, Dickerhoof, R.: 2004. ‘The medium is the message:
The costs and benefits of thinking, writing, and talking about life’s triumphs
and defeats’, Manuscript submitted for publication.
McCrae, R.R. and P.T. Costa: 1994, ÔThe stability of personality: Observations
and evaluationsÕ, Current Directions in Psychological Science 3, pp.
173175.
McCullough, M.E., Pargament, K.I. and Thoresen C.E., (eds): 2000, For-
giveness: Theory, Research, and Practice (New York: The Guilford Press).
McGregor, I. and B.R. Little: 1998, ÔPersonal projects, happiness, and mean-
ing: On doing well and being yourselfÕ, Journal of Personality and Social
Psychology 74, pp. 494512.
Nunnally, J.C. and I.H. Bernstein: 1994, Psychometric Theory (3rd ed.)
(McGraw-Hill, New York).
Oishi, S., E.F. Diener, R.E. Lucas and E.M. Suh: 1999, ÔCross-cultural vari-
ations in predictors of life satisfaction: Perspectives from needs and valuesÕ,
Personality and Social Psychology Bulletin 25, pp. 980990.
Parducci, A.: 1995, Happiness, Pleasure, and Judgment: The Contextual
Theory and its Applications (Erlbaum, Hove, England).
SUSTAINABLE HAPPINESS 85
Russell, B.: 1975, The Conquest of Happiness (London: Unwin original
work published 1930).
Ryff, C.D. and C.L.M. Keyes: 1995, ÔThe structure of psychological well-being
revisitedÕ, Journal of Personality and Social Psychology 69, pp. 719727.
Schulz, R.: 1976, ÔEffects of control and predictability on the physical and
psychological well-being of the institutionalized agedÕ, Journal of Personality
and Social Psychology 33, pp. 563573.
Scitovsky, T.: 1976, The Joyless Economy: The Psychology of Human Satis-
faction (Oxford University Press, New York).
Seligman, M.E.P.: 1991, Learned Optimism (A. A. Knopf, New York).
Sheldon, K.M. and A.J. Elliot: 1999, ÔGoal striving, need-satisfaction, and
longitudinal well-being: The Self-Concordance ModelÕ, Journal of Person-
ality and Social Psychology 76, pp. 482497.
Sheldon, K.M. and L. Houser-Marko: 2001, ÔSelf-concordance, goal-attain-
ment, and the pursuit of happiness: Can there be an upward spiral?Õ, Journal
of Personality and Social Psychology 80, pp. 152165.
Sheldon, K.M. and T. Kasser: 1995, ÔCoherence and congruence: Two aspects
of personality integrationÕ, Journal of Personality and Social Psychology 68,
pp. 531543.
Sheldon, K.M., T. Kasser, K. Smith and T. Share: 2002, ÔPersonal goals and
psychological growth: Testing an intervention to enhance goal-attainment
and personality integrationÕ, Journal of Personality 70, pp. 531.
Taylor, S.E. and J.D. Brown: 1988, ÔIllusion and well-being: A social psy-
chological perspective on mental healthÕ, Psychological Bulletin 103,
pp. 193210.
Watson, D., A. Tellegen and L. Clark: 1988, ÔDevelopment and validation of
brief measures of positive and negative affect: The PANAS scalesÕ, Journal
of Personality and Social Psychology 54, pp. 10631070.
Address for correspondence:
KENNON M. SHELDON
University of Missouri
65211 MO
Columbia
E-mail: SheldonK@missouri.edu
SONJA LYUBOMIRSKY
University of California
Riverside CA 92521
USA
E-mail: Sonja@cirtrus.ucr.edu
KENNON M.SHELDON AND SONJA LYUBOMIRSKY
86