UC Riverside
UC Riverside Previously Published Works
Title
Becoming happier takes both a will and a proper way: an experimental longitudinal
intervention to boost well-being.
Permalink
https://escholarship.org/uc/item/2gh028t2
Journal
Emotion (Washington, D.C.), 11(2)
ISSN
1528-3542
Authors
Lyubomirsky, Sonja
Dickerhoof, Rene
Boehm, Julia K
et al.
Publication Date
2011-04-01
DOI
10.1037/a0022575
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
Becoming Happier Takes Both a Will and a Proper Way:
An Experimental Longitudinal Intervention To Boost Well-Being
Sonja Lyubomirsky, Rene Dickerhoof, and
Julia K. Boehm
University of California, Riverside
Kennon M. Sheldon
University of Missouri, Columbia
An 8-month-long experimental study examined the immediate and longer term effects of regularly
practicing two assigned positive activities (expressing optimism and gratitude) on well-being. More
important, this intervention allowed us to explore the impact of two metafactors that are likely to
influence the success of any positive activity: whether one self-selects into the study knowing that it is
about increasing happiness and whether one invests effort into the activity over time. Our results indicate
that initial self-selection makes a difference, but only in the two positive activity conditions, not the
control, and that continued effort also makes a difference, but, again, only in the treatment conditions.
We conclude that happiness interventions are more than just placebos, but that they are most successful
when participants know about, endorse, and commit to the intervention.
Keywords: happiness, intervention, gratitude, optimism, effort, motivation
The pursuit of happiness and fulfillment is a goal shared by the
majority of people in the West and, increasingly, around the world
(Diener, 2000; Diener, Suh, Smith, & Shao, 1995). Historically
speaking, because of psychology’s prevailing focus on the allevi-
ation of psychopathology and weakness, relatively few studies
(e.g., see Boehm & Lyubomirsky, 2009; Lyubomirsky, 2008;
Lyubomirsky, Sheldon, & Schkade, 2005, for reviews) have actu-
ally attempted to increase individuals’ well-being (defined as high
life satisfaction, frequent positive affect, and infrequent negative
affect; Diener, Suh, Lucas, & Smith, 1999). The aim of the current
research was to test predictions from a model of well-being change
(Lyubomirsky, Sheldon, et al., 2005; Sheldon & Lyubomirsky,
2004), which argues that the intentional pursuit of greater happi-
ness—when done under optimal conditions—can be successful.
We began by testing two activities that have previously been
found to be effective in enhancing happiness—practicing optimis-
tic thinking by visualizing one’s best possible future selves (cf.
King, 2001) and expressing gratitude through writing (cf. Emmons
& McCullough, 2003). More important, however, we examined
two contextual or boundary conditions predicted by our model to
impact the efficacy of these activities—namely, (a) the intention to
use the intervention to become happier, as operationalized by a
self-selection factor, and (b) ongoing effort exerted toward the
activities specified by the intervention, as operationalized by ob-
jective ratings. For a framework to undergird these factors, we turn
to a model of the architecture of well-being change.
A Model of Well-Being Change
In their model of happiness, Lyubomirsky, Sheldon, and Sch-
kade (2005) proposed that three major factors contribute to peo-
ple’s levels of well-being: (a) their happiness set point (i.e., the
genetically determined stable level of happiness, which has been
shown to account for approximately 50% of the variance in indi-
vidual differences in well-being), (b) their life circumstances (e.g.,
factors such as income, marital status, or religiosity, which are
typically found to account for roughly 10% of individual differ-
ences in well-being), and (c) positive cognitive, behavioral, and
goal-based activities (which have the potential to account for a
significant portion, up to 40%, of individual differences in well-
being) (see also Sheldon & Lyubomirsky, 2004, 2006a, 2007).
This last factor, positive activity, has been the linchpin of recent
research efforts to bolster people’s happiness levels and served as
the focus of the current studies. Thus, we focused on positive
activities in the current research, examining factors that may
moderate the impact of such activities on well-being.
Although research on the heritability of well-being (e.g., Lykken
& Tellegen, 1996), the stability of personality (e.g., McCrae &
Costa, 1990), and the phenomenon of hedonic adaptation (Fred-
erick & Loewenstein, 1999) suggests that it should be very diffi-
cult, if not impossible, to become a lastingly happier person (see
Lyubomirsky, 2011; Lyubomirsky, Sheldon et al., 2005; Sheldon
& Lyubomirsky, 2004, 2006a, for reviews), evidence from earlier
and more recent experimental intervention studies suggests that
well-being can indeed be increased, at least in the short term (e.g.,
Sonja Lyubomirsky, Rene Dickerhoof, and Julia K. Boehm, Department
of Psychology, University of California, Riverside; Kennon M. Sheldon,
Department of Psychological Sciences, University of Missouri, Columbia.
Rene Dickerhoof is now with ICON Clinical Research, San Francisco,
California. Julia K. Boehm is now with the Harvard School of Public
Health.
The authors thank Ligia Ceja, Adrienne Grant, Kimberly Hazelwood,
Thomas Martin, Lisa McMoran, Sapna Mendon, Marianna Oganesyan,
Valerie Ozaki, and Anahit Sinoryan for assisting with data coding, partic-
ipant recruitment, and other aspects of this research. The authors are also
grateful to Joanne Frattaroli, Howard Friedman, and Robert Rosenthal for
valuable feedback and statistical assistance. This research was supported
by National Institutes of Mental Health R01 Grant 5 R01 MH065937-04.
Correspondence concerning this article should be addressed to Sonja
Lyubomirsky, Department of Psychology, University of California, River-
side, CA 92521. E-mail: sonja.lyubomirsky@ucr.edu
Emotion © 2011 American Psychological Association
2011, Vol. 11, No. 2, 391–402 1528-3542/11/$12.00 DOI: 10.1037/a0022575
391
Fordyce, 1977, 1983), and possibly over longer periods of time
(e.g., Seligman, Steen, Park, & Peterson, 2005). In other words,
although genes and personality traits may operate to hold happi-
ness levels relatively constant over time, and although individuals
may be predisposed to adapt to positive life experiences relatively
swiftly, engaging in happiness-increasing activities (such as com-
mitting to important goals, meditating, acting kindly toward others,
thinking optimistically, or expressing gratitude) has the potential to
improve levels of happiness for significant periods of time (Ly-
ubomirsky, Sheldon, et al., 2005; cf. Heller, Watson, & Ilies,
2004). Further, unlike many life circumstances and experiences to
which one adapts rather quickly (e.g., moving to a desirable new
home, office, or city), the practice of volitional positive activities
may serve as a natural antidote to the process of hedonic adapta-
tion (Lyubomirsky, 2011). That is, these activities, when practiced
using optimal timing and variation (see Sheldon, Boehm, & Ly-
ubomirsky, in press), may provide lasting boosts in well-being that
do not wear off entirely with the passage of time. To this end, the
aim of the current studies was to examine whether, when, and how
two such positive activities (expressing optimism and gratitude)
lead to improvements in well-being over time.
Interventions to Increase Well-Being via
Positive Activities
The first investigations to demonstrate that well-being can be
sustainably increased by engaging in happiness-increasing activi-
ties were conducted as far back as the 1970s and 1980s (Fordyce,
1977, 1983). In a series of studies, Fordyce sought to improve
well-being by teaching students in his classes what happy people
do and then have these students practice those behaviors in their
everyday lives. Results showed that students assigned to practice
14 different happiness-enhancing activities (e.g., spend more time
socializing, become more active, and develop optimistic thinking)
once daily over a 6-week period reported improved well-being for
more than 2 months, relative to a comparison group.
To directly test the hypothesis that intentional positive activ-
ity can boost happiness, Lyubomirsky, Sheldon, and colleagues
(2005) conducted two 6-week-long interventions that isolated two
specific activities (one behavioral and one cognitive) thought to
enhance well-being, committing acts of kindness and “counting
one’s blessings.” On the basis of research showing a positive
association between self-reported helping and self-reported happi-
ness (Dunn, Aknin, & Norton, 2008; Lyubomirsky, King, & Die-
ner, 2005; Sheldon et al., in press), the first experiment prompted
participants to engage in five acts of kindness on a particular
weekday (e.g., donating blood, feeding a friend’s pet). Results
showed that well-being increased for these participants, compared
to a control group (see also Sheldon et al., in press). The second
6-week intervention was based on previous research demonstrating
that “counting your blessings” can bolster feelings of satisfaction
and well-being (Emmons & McCullough, 2003; cf. Koo, Algoe,
Wilson, & Gilbert, 2008). In this study, participants were in-
structed to regularly think about things for which they were grate-
ful (e.g., “a healthy body,” “my mother,” and “having a Valen-
tine”). Relative to controls, participants who expressed gratitude
once a week (but not three times a week) indicated greater well-
being.
Seligman and colleagues (2005) tested the ability of five addi-
tional positive exercises (practiced over 1 week) to increase hap-
piness among an online sample of self-selected community-
residing participants. The results of this study showed that
participants who used personal strengths in novel ways each day
and those who considered three good things that happened to them
each day showed increases in their happiness and declines in their
depressive symptoms over a 6-month period. Indeed, these activ-
ities seemed to create ever-increasing happiness gains across
nearly all follow-up assessments. Participants who wrote and de-
livered a letter of gratitude also reported increased happiness and
reduced depression.
Besides such work with nonclinical samples, researchers have
also shown that positive psychological practices can relieve symp-
toms in mild to moderately depressed individuals, as well as in
outpatients with major depressive disorder and other affective
disorders (Fava, Rafanelli, Cazzaro, Conti, & Grandi, 1998; Fava
et al., 2005; Seligman, Rashid, & Parks, 2006).
Under What Conditions Do Positive Activities
Work Best?
In sum, a review of the small, but growing, well-being inter-
vention literature suggests that a number of positive activities can
increase well-being over both short and, possibly, longer term
periods of time. It is worth noting that well-being is conceptualized
here, and in previous work, in the hedonic (vs. eudaimonic) sense,
although the distinction has not been shown to be well-specified or
empirically supported (Kashdan, Biswas-Diener, & King, 2008).
However, our primary aim was not to test the efficacy of particular
positive activities or exercises per se, but rather to examine two
boundary conditions that might enhance or limit any activity’s
utility. The model of well-being change (Lyubomirsky, Sheldon, et
al., 2005; Sheldon & Lyubomirsky, 2004) proposes a number of
variables that should impact the efficacy of any positive activity,
not just the two activities investigated here.
Self-Selection
To illustrate the first boundary condition tested in the present
research, consider that the strongest happiness intervention results
have been reported by Seligman et al. (2005), who demonstrated
that several happiness-enhancing activities can maintain well-
being improvements for up to 6 months. In these studies, partici-
pants were fully aware of the study’s aim and had volunteered with
the hope (or expectation) that their well-being would increase. In
other words, their experiment used a self-selected sample of par-
ticipants who sought out a Web site about happiness and chose to
participate in an online study to improve well-being. Similarly, all
of Fordyce’s (1977) participants were self-selected and aware of
the purpose of the activities. In contrast, all the happiness-
enhancing interventions conducted by Lyubomirsky, Sheldon, and
colleagues to date (Lyubomirsky, Sheldon et al., 2005; Sheldon &
Lyubomirsky, 2006b; Sheldon et al., in press) have used designs in
which participants are unaware of the true purpose of the study.
That is, participants practice “cognitive exercises” and were not
informed that changes in happiness were our major focus. All of
these studies have yielded somewhat weaker or less durable effects
than the Seligman and Fordyce studies. This raises the possibility
392
LYUBOMIRSKY, DICKERHOOF, BOEHM, AND SHELDON
that a potentially important moderator of a happiness interven-
tion’s effectiveness is conscious knowledge of the intervention’s
purpose and motivation to help achieve that purpose.
When people purchase a self-help book, they are acting know-
ingly—choosing an experience that they hope will help them to
become a happier or better person. One might say that they are
“motivated” to engage in the experience, with some expectations
regarding the positive results that it might bring. Because of such
expectancies, later outcomes of the experience may be due to
motivated cognition—that is, demand effects, self-fulfilling proph-
ecy effects, or simple positive response biases (Kunda, 1990). For
example, when somebody praises the “Seven Steps to Happiness”
book that she just read, avowing that it really did make her happier,
how can one be sure she is not just caught in a positive illusion, or
trying to save face, or attempting to reduce dissonance? Obviously,
it would help to track her actual gains in happiness over time, as
observing tangible gains would at least be consistent with her
belief. However, we suggest that it would also help to compare
individuals like her with two other groups: (a) those assigned to
read the same happiness book who did not seek it out themselves
and (b) those assigned to read a different happiness book that
contains no effective advice (i.e., essentially a placebo). This
design would allow us to disentangle the effects of self-selection
(yes vs. no) from the effects of the substantive content of the
activity (treatment vs. control).
Considering activity content leads us to the notion that not all
happiness interventions will work, just as not all self-help books
will be helpful. The ideal self-help book would contain advice
supported by data, so that not only would participants be motivated
to act on it, but that they would also be doing the “right” thing (i.e.,
an activity that should work if practiced diligently and correctly).
Similarly, the ideal happiness intervention would contain practices
supported by empirical research. In sum, just believing one will
become happier is probably not enough: One also needs to be
doing something that is objectively effective at raising happiness.
In short, we argue that people need both a “will” and a “proper
way” to gain maximal benefits from a happiness intervention.
Analogously, medical patients need to know what their treatment
is and be motivated to follow it, but also must have been prescribed
the “right” treatment, not just a placebo. To be sure, a procedure
involving self-selection is usually avoided in experimental re-
search, because those who select themselves into a study expecting
benefits may systematically differ from those who engage in the
same study without such expectations. However, we believe it is
appropriate to examine the role of self-selection in the context of
a longitudinal happiness intervention, as this mirrors the goals and
experience of any consumer of positive psychology programs (or
self-help books).
Continued Effortful Practice of the Activity
Also consistent with the medical model, participants assigned to
a happiness intervention should gain the greatest benefit if they try
hard to carry it out over time. In other words, it is not enough to
knowingly choose the activity initially; one must also put objective
sustained effort into its practice. Consistent with this idea, Selig-
man et al. (2005) provide evidence that effortful pursuit of
happiness-enhancing activities is essential to their success. Specif-
ically, their study showed that continued adherence to the
happiness-enhancing activities during the course of the interven-
tion impacted the activities’ beneficial effects. Similarly, Sheldon
and Lyubomirsky (2006b) found that continued effortful perfor-
mance of an “optimistic thinking” activity predicted greater main-
tained change in subjective well-being, compared to a control
condition. Accordingly, we predicted that participants who exert
more effort during the weeks of practicing their assigned activities
would accrue greater benefit from them. However, the benefit
should only be observed if the activity is an efficacious one, which
provides the actual means for increases in well-being to accrue.
Based on this reasoning, we expected the biggest boosts in well-
being to be shown by participants who keep putting effort into the
activity and who are assigned to an “active” treatment condition, as
compared to an “inert” control condition.
The Present Study
Our intervention focused on two activities that have previously
been shown to bolster well-being for up to a minimum of 2
months—practicing optimistic thinking by visualizing one’s best
possible future selves (cf. King, 2001; Sheldon & Lyubomirsky,
2006b) and expressing appreciation toward others via gratitude
letters (cf. Emmons & McCullough, 2003; Lyubomirsky, Sheldon,
et al., 2005; Seligman et al., 2005). Accordingly, our study ran-
domly assigned participants to express optimism, express grati-
tude, or generate a list of the previous week’s activities (i.e., the
comparison control group; Lumley & Provenzano, 2003) once a
week over the course of an 8-week period. Measures of well-being
were administered immediately before, immediately after, and 6
months subsequent to the completion of the intervention period.
Our first hypothesis in this study concerned the overall efficacy
(i.e., main effect) of our two positive exercises. We predicted that,
relative to the control task, practicing either optimism or gratitude
would lead to improved well-being outcomes by the end of the
intervention, and that these improvements might even be main-
tained at the 6-month follow-up—although likely to a much lesser
degree. Notably, this main effect hypothesis was not essential to
the primary aims of this article, as we expected any main effects to
be qualified by the higher order boundary conditions that were the
true focus of this research.
Our second hypothesis concerned whether motivation to be-
come happier plays a role in well-being change. We predicted that
students who self-selected themselves into a so-called “happiness
intervention” would report greater gains in well-being relative to
students who self-selected into a “cognitive exercises” study. This
pattern was expected to be observed at both postintervention
assessment periods, although, again, we expected this main effect
hypothesis to be qualified, as gains may require both “a will” and
“a proper way.”
To address this essential issue, we also tested self-selection (i.e.,
motivation to become happier) as a moderator of the effects of our
two happiness-enhancing activities on increased well-being (Hy-
pothesis 3). Specifically, we predicted that the greatest improve-
ment in well-being (both postintervention and 6 months later)
would occur for “motivated” students instructed to express opti-
mism and gratitude, followed by “nonmotivated” students who
practiced these two activities, and, last, by students who wrote
about their past 7 days (i.e., both motivated and nonmotivated
individuals in the control condition). In other words, having the
393
OPTIMAL CONDITIONS FOR HAPPINESS-ENHANCING INTERVENTIONS
motivation and intention to pursue happiness is not enough; the
positive activity has to be appropriate and worthwhile. This would
be an important finding, indicating that happiness-intervention
effects are not mere self-selection effects; if control participants
who believe and hope that the activity will make them happier do
not actually become happier, then this would indicate that the
content of the activity matters, and that researchers are justified in
identifying and comparing happiness-enhancing practices.
Our last two hypotheses concerned the main and moderating
effects of continued effort on the impact of our two happiness-
enhancing activities on well-being. Similar to Hypothesis 2, con-
cerning the main effect of initial self-selection into the study,
Hypothesis 4 addressed the main effect of continued effort once
the study began. That is, those participants who exerted objectively
more effort in their assigned activity were expected to demonstrate
greater boosts in their well-being, compared with those partici-
pants who exerted less effort. More important, Hypothesis 5 pre-
dicted that the effort effect would be strongest in the two experi-
mental conditions and weakest or nonexistent in the control
condition.
Method
Participants
Undergraduate students (N ⫽ 355; 248 female, 107 male) at a
public university took part in this study. All students participated
in exchange for course credit, as well as up to $40 additional
compensation for the extra time the study required, including
completing it up to Time 2 (T
2
; immediate postintervention) and
Time 3 (T
3
; 6-month follow-up).
1
It should be noted that the
follow-up occurred when students had completed the relevant
course or graduated.
Twenty-three individuals who failed to complete at least four of
the eight weekly intervention assignments were removed from the
sample. In addition, two students were excluded as outliers whose
baseline well-being was more than 3 SDs from the sample mean,
leaving a final sample of 330 individuals (235 female, 95 male). Of
the 330 participants used in the final analysis, 13 failed to complete
T
2
(3.9%) and 120 failed to complete T
3
(36%). The T
3
attrition
rate is comparable to the average attrition rate (42%) found in a
meta-analytic review of 152 longitudinal studies, including 48 that
used college samples (Roberts & DelVecchio, 2000).
Over one-third of the sample was of Asian descent (40%), 20%
were Hispanic, 17% were Caucasian, 5% were African American,
5% were Hawaiian/Pacific Islander, 6% indicated “more than one
ethnicity,” and 7% identified themselves as “other.” Students
ranged in age from 18 to 46 years (M ⫽ 19.66, SD ⫽ 2.91).
Design
A 2 (Self-Selection: Yes vs. No) ⫻ 3 (Activity: Optimism,
Gratitude, Control) factorial design was used. Potential partici-
pants were offered the option to take part in one of two posted
studies: one advertised as a happiness intervention and the other as
a study involving cognitive exercises. Ample slots were available
for each “study” at all times. In this way, students self-selected into
either a “low motivation” group (those who signed up for the
cognitive exercises study and who, presumably, were not as in-
vested in becoming happier; n ⫽ 99) and a “high motivation”
group (those who signed up for the happiness intervention and,
presumably, desired to be happier; n ⫽ 231). All participants—
regardless of the study they signed up for—were then randomly
assigned to express optimism (n ⫽ 112), convey gratitude (n ⫽
108), or generate a list of their experiences over the past week (n ⫽
110; i.e., control).
Procedure
Introductory appointment. Except for an introductory face-
to-face appointment that took place in a laboratory, this study was
conducted entirely over the Internet, using a Web site accessible
only to our participants.
2
At this appointment, groups of 5–10
participants received a verbal description of the study. Notably,
regardless of the “study” they signed up for (i.e., both low- and
high-motivation groups) all participants were told that the aim of
the study was to improve well-being. This information was pre-
sented at this time in accord with procedures of standard placebo-
controlled designs, in which all participants are informed about the
“treatment” that is being tested, which is said to have a reasonable
chance of working (Rosenthal, 1976). Thus, we could test whether
the experimental conditions would differ from the control condi-
tion, which should provide evidence for a placebo-type effect (if
any). It is also noteworthy that this design isolates the effects of the
self-selection factor, because all participants’ expectancies of the
study’s aim are equated once they arrive at the lab, and all that
differs is the experience they initially thought they would receive.
Baseline assessments. After receiving their instructions, stu-
dents accessed the study’s Web site and answered the first set of
questionnaires at their convenience. This first baseline assessment
consisted of a consent form, demographic questions, and measures
of the students’ well-being. Approximately 1 week later, partici-
pants completed a second baseline assessment by again accessing
the study’s Web site at their own discretion and responding to
identical questions assessing the same outcomes. To establish a
relatively more stable baseline for these measures, we combined
ratings from the first and second baseline assessments of each of
the four primary outcome variables—pleasant affect, unpleasant
affect, life satisfaction, and happiness. Test-retest correlations for
these variables between the two assessments ranged from .55 for
unpleasant affect to .84 for happiness. This stable baseline (re-
ferred to as T
1
) was used to minimize measurement error that can
occur with single-administration measures of well-being (e.g., the
transient effects of weather and/or recent events; Schwarz & Clore,
1
Due to space constraints, we do not present well-being data from an
additional 9-month follow-up. The pattern of results was similar to previ-
ous time periods, albeit weaker and sometimes marginally significant or
nonsignificant. Also, in the interest of space, data on several other variables
collected in this study are not described; they are available from the first
author.
2
Online data collection has been shown to reduce the likelihood that
participants will misrepresent their actual behaviors and to increase the
likelihood that they will reveal sensitive and personal information (Turner
et al., 1998; Velicer, Prochaska, Rossi, & Snow, 1992). This methodology
was thought to be particularly beneficial in the current study, in which all
dependent variables are self-report.
394
LYUBOMIRSKY, DICKERHOOF, BOEHM, AND SHELDON
