Do happy people care about society’s problems?

Abstract

Since ancient times, scholars, individuals, and societies have been preoccupied with the pursuit of happiness. But might individual happiness actually be bad for society and the world? A common concern – which we refer to as the Pollyanna hypothesis – is that happy people might be too happy to care enough about important current issues, thus being less likely to act on improving society and the world. In three studies, however, we found that feeling good predicted more, not less, action on current issues. We saw this pattern in the context of the 2017 far right rallies in Charlottesville, VA (Study 1), a wide range of social, political, and environmental issues chosen by participants (Study 2), and environmental action within a nationally representative sample (Study 3). These correlational findings speak against the Pollyanna hypothesis: Happiness does not seem to preclude caring about local and global issues.

Full text

Do happy people care about society’s problems?
Kostadin Kushlev
a
, Danielle M. Drummond
b
, Samantha J. Heintzelman
c
and Ed Diener
d
a
Georgetown University, Washington, DC, USA��;
b
University of Utah, Salt Lake City, UT, USA;
c
Rutgers University, Newark, NJ, USA;
d
University
of Virginia, Charlottesville, VA, USA
ABSTRACT
Since ancient times, scholars, individuals, and societies have been preoccupied with the pursuit
of happiness. But might individual happiness actually be bad for society and the world?
A common concern – which we refer to as the Pollyanna hypothesis – is that happy people
might be too happy to care enough about important current issues, thus being less likely to act
on improving society and the world. In three studies, however, we found that feeling good
predicted more, not less, action on current issues. We saw this pattern in the context of the 2017
far right rallies in Charlottesville, VA (Study 1), a wide range of social, political, and environmental
issues chosen by participants (Study 2), and environmental action within a nationally representa-
tive sample (Study 3). These correlational ndings speak against the Pollyanna hypothesis:
Happiness does not seem to preclude caring about local and global issues.
ARTICLE HISTORY
Received 22 May 2019
Accepted 22 June 2019
KEYWORDS
Happiness;
helping/pro-social behavior;
well-being; emotion�
In 2011, the United Nations dened happiness as
a ‘fundamental human goal’ and invited Member
States to pursue measures that enhance the happiness
of their citizens. As nations across the globe become
more interested in human happiness, some have
expressed concerns about the downsides of being hap-
pier and the so-called eld of ‘positive psychology’
(Bohart & Greening, 2001; Lazarus, 2003). What if in
our rush to make everyone happy, people became
complacent about the plight of their local communities,
society, and the world?
The power of negative thinking
Perhaps precisely in reaction to the expanding focus on
happiness and well-being, a growing number of popular
books (Kashdan & Biswas-Diener, 2015; Knight, 2014) and
articles (Adler, 2018; LaBier, 2017) have argued that we
should, instead, be embracing our darker side, touting
the ‘power of negative thinking’ and emotions. Such pop-
ular critiques of positivity seem to be fundamentally based
on a functional perspective of emotion – the idea that
(negative) emotions exist to solve particular problems
within one’s environment (Ekman, 1992). This functional
perspective underlies both evolutionary and constructivist
theories of emotion and is well supported by research
(Keltner & Lerner, 2010). For example, anger has been
associated with intentions to take action, particularly in
response to perceived unfairness or wrongdoing by
transgressors (Roseman, Wiest, & Swartz, 1994). In one
study, British citizens’ anger towards the British govern-
ment predicted not only desire to punish those responsible
for the war in Iraq, but also intentions to advocate with-
drawal from Iraq and to support compensation to the Iraqi
people (Iyer, Schmader, & Lickel, 2007). Even shame – an
emotion that is largely seen as maladaptive in the psycho-
logical literature (Tangney, Stuewig, & Mashek, 2007) – has
recently been shown to promote (at least) the desire for
action in the intergroup context (Iyer et al., 2007).
Going beyond specic negative emotions, negative
moods seems to also confer benets, including improved
judgment accuracy, reduced gullibility, greater persever-
ance, and more politeness (for a review, see Forgas, 2013).
According to the feelings-as-information perspective
(Schwarz, 2012), people often use their existing mood
state to form a judgment of how they feel about external
events. In the context of social, political, or environmental
issues, a person prone to negative moods may, therefore,
decide they feel more concerned about such issues than
would those in more positive moods. Indeed, negative
aectivity – the general tendency to experience negative
emotions as a personality trait – has also been shown to
confer benets for both the detection and action in the
face of danger. A meta-analysis of 75 studies, for example,
showed that depressed individuals were more realistic
about the future than nondepressed individuals
(Oettingen, 2000). And in a lab study, groups of three
participants, who were led to believe that there was
CONTACT Kostadin Kushlev kk1199@georgetown.edu
Materials and data are available on the Open Science Framework: https://osf.io/86ngd/?view_only=70ef0a3624ed427aaccf308428530a36
THE JOURNAL OF POSITIVE PSYCHOLOGY
https://doi.org/10.1080/17439760.2019.1639797
© 2019 Informa UK Limited, trading as Taylor & Francis Group

a computer re in the lab, were faster at reacting to the
perceived danger when at least one member of the group
had high neuroticism (Ein-Dor, Mikulincer, & Shaver,
2011). Thus, by perceiving and responding to danger,
people high in negative aect may benet themselves –
as well as those around them.
For what it’s worth, the empirical evidence seems to t
with a stereotype of the angry social activist: somebody
who joins rallies, argues with those who do not share their
point of view, or composes negative tweets. Though
direct evidence is lacking, it is reasonable to hypothesize
that negative aect may be associated with both greater
concern and more action in response to current social
issues and environmental threats.
A darker side of positivity
Just as negativity may predict greater propensity to act,
too much positivity may predict less action in response to
current or anticipated problems. The National Cancer
Institute, for example, estimates that people who under-
estimate their risk of heart disease – that is, unrealistic
optimists – tend to develop earlier signs of cardiovascular
problems (Ferrer et al., 2012). Similarly, individuals overly
optimistic about health risks were less likely to engage in
health-protective behavior, such as buying radon test kits
after being advised of the possible presence of this radio-
active element (Weinstein & Lyon, 1999). Beyond under-
estimating health risks, unrealistic optimism about one’s
own future life more generally may also be detrimental: In
a longitudinal study of almost 7,000 German residents,
older adults who overestimated how satised they would
be with their lives ve years into the future were more
likely to die over a 12-year period (Lang, Weiss, Gerstorf, &
Wagner, 2013). These striking eects of unrealistic opti-
mism may be due to a range of maladaptive tendencies
that accompany it – from using defensive strategies when
faced with information about health risks to actually
engaging in riskier behaviors (for a review, see Chang,
2008; Forgas, 2014).
Beyond the domain of personal health, preliminary
evidence suggests that very happy people might be
less engaged in political action. In a study of over one
hundred thousand individuals across 96 nations, people
rating their satisfaction with life as 10 out of 10 were
less likely to engage in behaviors, such as signing
a petition or joining a protest, compared to individuals
rating their lives 8 or 9 (Oishi, Diener, & Lucas, 2007).
The benets of subjective well-being
Most of the evidence for a negative association
between positivity and action is based on people with
overly optimistic expectations about the future or with
extreme satisfaction with their lives. Research on sub-
jective well-being and its components – high life satis-
faction, frequent positive aect, and infrequent
negative aect – suggests that being happy can be
benecial in a range of domains. Happier people, for
example, are healthier (Pressman & Cohen, 2005) and
live longer (Diener & Chan, 2011). These eects are
produced at least in part because happy people engage
in more healthy behaviors (Boehm, Vie, & Kubzansky,
2012), taking preventative action to mitigate risk (Kim,
Kubzansky, & Smith, 2015), and avoiding risky behaviors
like not using sun protection (Grant, Wardle, & Steptoe,
2009).
Beyond taking better care of themselves, happy peo-
ple might be more caring and helpful to others. Happy
employees, for example, help other workers and the
company more than their less happy counterparts,
even when the help they are giving is not part of
their job description (Boehm & Lyubomirsky, 2008).
Though such organizational citizenship is not equiva-
lent to societal citizenship, these ndings suggest that
happy people may be more, rather than less, inclined to
engage in action that benets society. Indeed, happier
people are more likely to donate money to charity and
more likely to volunteer (Lyubomirsky, King, & Diener,
2005; Oishi et al., 2007). In the realm of political action,
happiness has also been associated with a greater like-
lihood to vote (Weitz-Shapiro & Winters, 2011).
The pollyanna hypothesis
Our review of the existing literature suggests that nega-
tive emotions seem to stir people to action, and so does
feeling good – except when it doesn’t. Past research
suggests that negative aect may prompt greater
action in response to various threats and stronger eort
to rectify perceived transgressions. But while overly
optimistic and extremely happy people are documen-
ted to engage in riskier behavior and participate less in
some kinds of political action, subjective well-being
generally has a positive association with healthy beha-
vior, interpersonal prosocial behavior, and political
activity like voting. This apparently conicting state of
the literature provides an unsatisfactory answer to the
concern – shared by popular and academic authors
alike – that eorts to make people ever happier may
be counterproductive, producing a society of
Pollyannas unconcerned with making progress on sol-
ving critical issues and threats faced by their commu-
nities, nations, and the world.
A falsiable test of the Pollyanna hypothesis requires
that we clearly dene happiness as the presence and
2K. KUSHLEV ET AL.

frequency of positive emotions – rather than as the
absence of bad emotions. In particular, the Pollyanna
hypothesis states that feeling good – feeling happy, joyful,
content – should preclude people from also feeling bad –
concerned, worried, angry, or guilty – about important
current issues. Thus, the Pollyanna hypothesis predicts
a negative relationship between the positive aect people
generally experience in their daily lives and the negative
aect people feel in response to specic events or issues.
Notably, the Pollyanna hypothesis does not necessarily
suggest that general negative aect is good for enacting
change on current issues, but simply that general positive
aect may decrease concern about current issues, which
is useful in motivating action on those issues. Note, how-
ever, that even though the Pollyanna hypothesis is
a causal hypothesis, falsiable evidence inconsistent
with that hypothesis does not have to be. Thus, the
Pollyanna hypothesis becomes less tenable if general
positive aect is not inversely related to negative aect
in response to specic issues, as well as with their past,
current, and future action on that issue. Of course, evi-
dence that positive aect predicts greater negative aect
and greater action in response to an issue would be even
more inconsistent with the Pollyanna prediction.
The present research
In Study 1, we explored the relationship of positive
aect with specic negative aect and action with
regards to The Unite the Right Rally in Charlottesville,
VA on August 11
th
and 12th, 2017. In Study 2, we
invited participants to identify their own ‘social, politi-
cal, or environmental issue’ that they felt was impor-
tant. We then again assessed the relationships between
general positive aect, specic negative aect, and
action. Finally, in Study 3, we used a broader, represen-
tative sample from the U.S. General Social Survey (GSS),
examining the association of happiness with environ-
mental concerns and behavior.
Study 1: the far-right rallies in charlottesville
In Study 1, we explored how levels of positive and negative
aect impacted students’ responses to a troubling local
event. In particular, we were able to access a sample of
students in Charlottesville, VA in the aftermath of Unite the
Right Rally – the far-right rally and ensuing counter protests
of August 11
th
and 12
th
, 2017 that garnered national atten-
tion. This study oered a unique opportunity to explore
how happy and unhappy individuals would act after their
surrounding community had been impacted by a tangible
emotional event, amidst national outcry and calls for
action.
Method
In total, 320 students enrolled in the study (Age: M = 18.8,
SD = 0.98; 52% Female). All students were recruited
through the University of Virginia participant pool for
class credit. The sample size for this time-sensitive survey
was determined by practical considerations: the research-
ers’ assigned credits for recruitment from the participant
pool in the semester following the rallies. Sensitivity
analyses indicated that this sample size gave us 80%
power to detect true population eects of size ρ= .16.
The most common race/ethnicity was non-Hispanic
White (67%), followed by Asian/Asian American (13%),
Black/African American (8%), and Hispanic/Latino (6%).
On social issues, rated from 1 (very conservative) to 10
(very liberal), participants tended to be more liberal
(M= 6.78, SD = 2.10), while on economic issues they
were relatively centrist on the political spectrum
(M= 5.2, SD = 2.23). Participants completed an online
survey through Qualtrics.
We operationalized happiness as the positive aect
people had experienced over the preceding month and
unhappiness as the negative aect people had experi-
enced over the same period. As positive and negative
aect are fairly independent (Watson, Clark, & Tellegen,
1988), these operationalizations of subjective well-being
best served the key purpose of this research to investigate
whether feeling good or feeling bad predicts more action
in response to perceived threats and issues. We thus asked
participants to rate their positive and negative aect on
the SPANE scale (Diener et al., 2010), reporting how much
they felt various emotions over the past four weeks on
a scale from 1 (not at all or very slightly) to 5 (very much or
extremely strongly). See Table 1 for descriptives.
1
Participants then read a short paragraph describing the
controversial Unite the Right Rally that took place 3 months
earlier. The description oered a brief summary of the
events, while refraining from providing interpretations of
the event (see materials on OSF). Participants were asked to
imagine they were explaining the event to a friend, focus-
ing on the thoughts and emotions they experienced.
Instructions prompted participants to keep responses to
a minimum of 5–7 sentences, but to limit their writing to no
more than 5 minutes. These tasks were intended to refresh
initial sentiments about the event and allow a more accu-
rate assessment of their event-specic aect.
As a measure of concern about the issue, we measured
participants’ negative aect regarding the Charlottesville
rallies. We thus asked participants to complete the SPANE
again, but with modied instructions about how they feel
when thinking about the rallies. To measure behavioral
intentions and past behavior, participants were given
a list of 10 possible actions and asked to indicate if they
THE JOURNAL OF POSITIVE PSYCHOLOGY 3

(1) have already done, (2) plan to do, or (3) have no inten-
tion to do the specied action. Options ranged from intel-
lectual actions (e.g. engage in meaningful conversation
with friends) to tangible actions (e.g. write a blog post,
join a protest, give money) in order to cover a full spectrum
of possible reactions. Lastly, to measure current willingness
to act, we presented participants with the option to ‘join
a community of peers’ available to support and assist in
instances of intolerance, instilling a sense of accountability
in their response. They recorded their interest in this group
on a continuum from −3 (very disinterested)to 3 (very
interested), recoded to a 1-to-7-point scale for analyses.
See Table 1 for descriptive statistics (materials and data
available: https://osf.io/86ngd/?view_only=
70ef0a3624ed427aaccf308428530a36).
Results
Concern
General positive aect expectedly predicted lower gen-
eral negative aect, r= −0.24, 95%CI [−.14; −.34],
p < .001; yet, general positive aect predicted margin-
ally greater negative aect about the rallies, r = .09,
95%CI [−.02; .20], p= .091. In contrast, despite being
measured with the same items, general negative aect
was not a strong predictor of specic negative aect
about the issue, r= .05, 95%CI [−.06; .16], p> .250.
Action
General positive aect did not predict past action, r= .06,
95%CI [−.05; .17], p> .250, but neither did general negative
aect, r= .01, 95%CI [−.10; .12], p> .250. We also found
that positive aect did not predict future intentions to act,
r= 0.02, 95%CI [−.09; .13], p> .250, but negative aect did,
r= 0.12, 95%CI [.01; .23], p= .029. Looking at current action
by measuring interest in joining a support peer group,
however, we found that general positive aect was
a signicant predictor of current action, r= 0.14, 95%CI
[.03; .25], p= 0.012, whereas negative aect was a margin-
ally signicant predictor of current action, r= 0.11, 95%CI
[.00; .22], p= .053.
Summary
Study 1 failed to produce evidence of the Pollyanna
hypothesis that happiness is associated with less action
in response to important social concerns. To the con-
trary, happier individuals, when presented with the
opportunity to do so, were more willing to act in the
present, helping the community cope with the after-
math of a traumatic local event.
Study 2: choose your own issue
In Study 2, we allowed participants recruited across the
United States to select the social, political, and environ-
mental issues that are most personally important to
them. This approach allowed us to examine the rela-
tionship between happiness and action across a wider
range of issues important to the American public.
Method
We recruited 544 participants (Age: M= 28.2; SD = 11.7),
241 of whom were recruited through the participant pool
of a public American university (M = 18.7, SD = 0.87) and
the other 303 were recruited through Mechanical Turk with
the help of TurkPrime (Age: M = 35.8, SD = 10.1; Modal
Education = Bachelor Degree [n = 125]). Sensitivity analyses
indicated that this sample size gave us 80% power to
detect true population eects of size ρ= .12. The most
Table 1. Descriptive statistics.
Study 1: Far Right Rallies in Charlottesville, VA (2017)
PA NA Concern Past Action Current Action Future Action
N 320 320 320 320 320 320
Mean 3.83 2.61 4.08 2.31 4.78 2.55
SD 0.77 0.90 0.84 1.48 1.64 2.40
Study 2: Self-Selected Issues
PA NA Concern Past Action Current Action Future Action
N 544 544 543 544 544 544
Mean 3.55 2.36 3.94 1.98 0.24 2.24
SD 0.92 0.93 0.91 1.62 0.43 1.77
Study 3: Environmental Action (GSS)
Happiness Concern Past Action Current Action Future Action
N 1428 1404 – 1423 544
Mean 2.11 3.86 – 2.22 2.24
SD 0.64 1.10 – 0.72 1.77
Note. PA = positive affect (general); NA = negative affect (general). As other constructs were assessed with a range of tools across studies, means
and standard deviations should not be directly compared across studies.
4K. KUSHLEV ET AL.

common race/ethnicity was non-Hispanic White (70%), fol-
lowed by Asian/Asian American (16%), Black/African
American (7%), and Latino(a)/Hispanic (4%). The median
religious aliation, on a scale from 1 (not at all religious) to
10 (extremely religious), was 4 (M= 4.41 SD = 3.02).
Participants completed an online survey on Qualtrics
(see Table 1 for descriptive statistics for all measures).
As in Study 1, they rated their positive and negative
aect on the SPANE (Diener et al., 2010), reporting how
they felt over the past four weeks on a scale from 1 (not
at all or very slightly) to 5 (very much or extremely
strongly). Participants were then asked to think about
and type in as many political, social, or environmental
issues that worried them. On the following page, parti-
cipants were invited to select the one issue of those
they had just listed, and to provide up to 3–5 reasons
why this issue was important to them. Participants most
frequently selected environmental issues (26.3%), fol-
lowed by concerns of the current state of the
U.S. government (11.8%), healthcare (8.8%), and racism
(6.6%). As we intended, however, groups of 2% to 5% of
participants also selected a variety of other issues
across the political spectrum, including the state of
the economy, wealth inequality, terrorism, global war,
gun violence, loss of gun rights, decline in morals,
humanitarian crises, women’s rights, LGBT rights, immi-
grant rights, and illegal immigration.
After selecting an issue they cared about, participants
were asked to recall and describe ‘a time or event when
[they] felt particularly worried about this issue.’ These
prompts were designed to ensure we assessed their con-
cern regarding the issue more accurately. As in Study 1, we
operationalized concern by asking participants to complete
the negative aect of the SPANE but with reference to how
they feel when thinking about the issue (e.g., angry, sad,
afraid). These items of negative aect about the issue
formed a composite measure of concern (α= .90).
To assess past and future behavior, we asked partici-
pants to indicate whether they had done or plan to do
each of six dierent actions. The given options were
similar to those in Study 1 but more limited in number
to ensure that they were applicable to the wider range of
issues examined in Study 2. The action options included
general ways to enact political and social change in
a democratic society (e.g., joining a protest/rally, contact-
ing their representative, giving money to organizations
whose mission is to solve/resolve the issue). As in Study 1,
we summed the number of actions people had done or
planned to do to form separate composites of past and
future behavior (varying from 0 to 6).
As a measure of current behavior, we asked partici-
pants whether they were interested in keeping informed
about the issue by signing up for an email newsletter; we
then coded whether or not participants provided their
email (1) or not (0), providing an objective measure of
behavior. Though typing in one’s email address in
a survey box does not represent, per se, a signicant eort,
we reasoned that the symbolic meaning of this action
represents a commitment to stay informed about the
issue on a regular basis. To quantify the level of commit-
ment that participants signied by providing their email,
we asked participants to indicate how frequently they
wanted to receive such newsletter with the anchors of: 0
(no emails), 1 (a few times a year), 2 (once a month), 3
(biweekly), 4 (once a week), 5 (several times a week), and 6
(daily). To make this commitment salient to participants,
this question came before participants were invited to
sign up for the newsletter. Out of the 129 participants
who provided their email (or approx. 1 in 4 of the sample
as shown in Table 1), the modal response (n= 42) was to
choose to receive the newsletter once a week; the median
response was ‘Biweekly’ (M= 2.23; SD = 1.75), and 8.5%
chose to receive an email every day.
As in Study 1, additional variables beyond the focus of
the present report were also included (see https://osf.io/
86ngd/?view_only=70ef0a3624ed427aaccf308428530a36
for materials and data).
Results
Concern
We found that general negative aect predicted signi-
cantly greater specic negative aect about the issue,
r= .09, 95%CI [.00; .17], p= .041. Positive aect, however,
also predicted higher, not lower, negative feelings about
the issue, r= .14, 95%CI [.06; .23], p< .001; in contrast,
general positive aect was inversely related to general
negative aect, r= −.44, 95%CI [–.50; −.36], p< .001.
Action
Positive aect predicted greater concern about an issue
people cared about, but does it also predict greater
propensity for action? We found that happy people
were more likely to have been proactive in addressing
the issue they were worried about, r= .09, 95%CI [.00;
.17], p= .040. At the same time, unhappy people –
those higher in general negative aect – were less likely
to have done much about the problem they chose as
important to them, r= −.12, 95%CI [−.20; −.03], p= .006.
Yet, the more specic negative aect participants felt
about the issue, the more likely they were to have done
more about the issue, r= .17, 95%CI [.09; .25], p< .001.
Indeed, specic negative aect mediated the relation-
ship between general positive aect and past action,
indirect eect = .04, 95%CI[.01; .09], z= 2.06, p= .039,
THE JOURNAL OF POSITIVE PSYCHOLOGY 5

explaining 26% of the total association between PA and
past action.
Next, we examined participants’ intentions to do
something about the issue in the future, as assessed
with the same list of possible actions as we used for
assessing past behavior. Positive aect had virtually no
bearing on how much people planned to do about the
issue in the future, r= −.01, 95%CI [−.09; .07], p> .250,
and the association of intended behavior with general
negative aect was also nonsignicant, r= .06, 95%CI
[−.03; .14], p= .189. Even specic negative aect about
the issue did not predict greater intention to engage in
future activity as measured in the present study, r= .02,
95%CI [−.06; .10], p> .150. Notably, past behavior was
a negative predictor of future behavior, r= −.25, 95%CI
[−.16; −.32], p< .001. This pattern suggests that, if
people had already sought to, for example, inform
themselves about the issue, they might have felt it
was inaccurate to also say that they plan to get
informed about the issue.
Finally, turning to our measure of current behavior,
a logistic regression indicated that even a single point
increase in positive aect was associated with 31%
greater likelihood to provide one’s email to sign up
for the newsletter, exp(B) = 1.31, z= 2.33, p= .020, χ
2
(1) = 5.69. For consistency in providing eect sizes, we
also estimated the Pearson correlation, yielding the
same conclusions, r= .10, 95%CI [.02; .18], p= .019. In
contrast, general negative aect was not associated
with the likelihood of signing up for the email news-
letter: exp(B) = 1.03, z = 0.30, χ
2
(1) = 0.09; r= .01, 95%CI
[−.07; .10]; ps > .250. As with future action, specic
negative aect also did not predict signicantly greater
likelihood of signing up, exp(B) = 1.19, χ
2
(1) = 2.23;
r= .06, 95%CI [−.02; .15], ps > .140. We did not nd
any associations with the frequency of wanting to
receive these emails.
Summary
Across a broader national sample of individuals who chose
a current issue that they cared and felt worried about, we
found further evidence against the possibility that happy
people are unconcerned about current issues or that they
are unwilling to be proactive in addressing these issues.
General positive aect predicted feeling greater negative
aect about issues people chose as most important to
them, having taken a greater number of past actions, and
being more willing to act when given the opportunity to
do so in the present (i.e. to provide one’s email to stay
informed about the issue through a regular newsletter).
Though general negative aect was also related to more
issue-specic negative aect, this general propensity to feel
negative emotions predicted having engaged in fewer past
actions and did not predict greater likelihood to act when
given the opportunity to do so. In contrast, issue-specic
negative aect was the strongest predictor of past beha-
vior (though not of current action or future intentions).
Thus, the evidence suggests that negative emotions are
not necessarily associated with less action; rather, although
general unhappiness in life may indeed predict less action,
situationally-bound negative emotions about a specic
issue may be adaptive, predicting more action on the
issues people are worried about.
Though Studies 1 and 2 provide consistent evidence,
the nonrepresentative samples do not allow us to
meaningfully examine the possibility that the observed
relationships may be explained by third variables, such
as common demographics. Richer individuals, for exam-
ple, might be both happier and have more resources to
dedicate to taking action on the issues that they care
about. Accordingly, in Study 3, we use nationally repre-
sentative data from the GSS to address these questions.
Study 3: environmental action (general social
survey: GSS)
In Study 3, we used a broader, representative sample
from the U.S. General Social Survey (GSS) in 2010 to
examine the role of happiness in the context of envir-
onmental issues – the category of issues most com-
monly chosen by participants in Study 2.
Method
In all, 2044 participants were polled in GSS 2010 (Median
age = 47, Range age = 18–89; 56% female; Median years of
education = 13). A representative subsample of people was
asked questions on the environmental module, which was
included in two out of three versions of the 2010 GSS poll
(Ballot A: n= 667, and Ballot B: n= 763). Thus, the total
number of participants with data on the environmental
variables was N= 1430. Sensitivity analyses indicated that
this sample size allowed us to detect true population
eects of size ρ= .07 with 80% power. As the GSS carefully
samples each subballot, the demographic prole of the
subsample used in the current study was identical to the
one reported above for the full GSS 2010. Of course, due to
small nonresponse rates, the exact ns vary slightly across
dierent variables (see Table 1 for details and for descrip-
tive statistics for all key variables). Participants in the GSS
indicate whether they are very happy, pretty happy, or not
too happy. As a measure of environmental concern, we
used responses to the item: ‘Generally speaking, how con-
cerned are you about environmental issues?’ assessed on
a scale ranging from 1 (not at all concerned) to 5 (very
concerned). Participants also responded to a more specic
6K. KUSHLEV ET AL.

item about climate change: ‘In general, do you think that
a rise in the world’s temperature caused by climate change
is dangerous to the environment?’ from 1 (not at all dan-
gerous) to 5 (extremely dangerous).
We found four items of current daily behaviors in the
GSS relevant to environmental issues, including recycling,
avoiding products that may hurt the environment,
attempting to reduce fuel consumption while driving,
and driving less altogether – all rated on scales of 1
(never), 2 (sometimes), 3 (often), and 4 (always), α = .69.
We identied three items in the GSS that measured peo-
ple’s willingness to sacrice in the future in order to
protect the environment, including a willingness to pay
higher taxes, willingness to pay higher prices, and will-
ingness to accept a cut in living standards (see Table 1 for
means and standard deviations). These items were rated
on continuous scales, from 1 (not at all willing) to 5 (very
willing), formed an internally consistent composite,
α = .84, and were analyzed together.
Results
Because happiness in the GSS is measured with three
clearly dened categories – rather than with a multi-item
continuous composite scales – we used ANOVAs to com-
pare mean dierences in concern and action between
people who are very happy, those who are pretty happy,
and those who are not too happy (see Figure 1). To test
whether the eects can be explained away by common
demographics, we add covariates using ANCOVAs (see
Table 2).
Concern
An ANOVA comparing not too happy, pretty happy, and
very happy people indicated no overall dierences in gen-
eral environmental concern, F(2, 1399) = 0.14, p> .250. An
ANOVA predicting concern about the climate, however, did
indicate an omnibus dierence between people of varying
levels of happiness, F(2, 1320) = 3.96, p= .019. This omnibus
eect was not explained by age, gender, income, and
education when added to an ANCOVA – even though
younger people (β
age
= −.10, p< .001), the less wealthy
(β
income
= −.08, p< .013), and females (β
sex
= .06, p< .029)
worried signicantly more about climate. Pairwise compar-
isons of the marginal means in the ANCOVA indicated that
‘not too happy’ individuals were more concerned about
the climate than ‘very happy’ individuals, p
tukey
= .033 (see
Table 2 for details).
Current behavior
Unhappy people were more likely to worry about climate,
but were they more likely to act to protect the environ-
ment? The ANOVA indicated an overall omnibus eect on
current behavior between people with diering happiness
levels, F(2, 1418) = 4.68, p= .009. Unlike the results for
climate concern, however, very happy people were most
likely to engage in environmentally friendly behavior
(Figure 1). In particular, pairwise comparisons indicated
that ‘very happy’ participants were more likely to engage
in daily environmental action than both ‘not too happy’
participants, MD(SE) = .14(.06), p = .024, and ‘pretty happy’
participants, MD(SE) = .13(.04), p = .004; in contrast, ‘not too
happy’ participants were not signicantly less likely to take
daily environmental action than ‘pretty happy’ participants,
MD(SE) = −.01(.06), p> .250. Despite these dierences in
the predictors of concern and action, people who were
more concerned about climate were, expectedly, more
likely to act to mitigate their environmental impact,
r= .38, p< .001. Older (β
age
= .09, p< .001), richer
(β
income
= .02, p< .001), and more educated people (β
educ
= .13, p< .001), as well as females (β
sex
= .06, p< .001), were
more likely to be environmentally friendly. Controlling for
these demographics, however, did not fully explain the
eect of being very happy on current environmental action
(Table 2).
Future sacrifice
The eects on willingness to make personal sacrices mir-
rored those on current action. We observed a marginally
signicant omnibus eect of happiness category, F(2,
Figure 1. Individuals who are not too happy worry most about climate, but very happy individuals are most likely to act to protect
the environment in the nationally representative sample in Study 3 (GSS).
THE JOURNAL OF POSITIVE PSYCHOLOGY 7

1395) = 2.88, p= .057, which was signicant after control-
ling for demographic factors (Table 2). Using pairwise com-
parisons, we again found no dierence between ‘not too
happy’ individuals and ‘pretty happy’ individuals,
MD(SE) = .01(.09), p> .250, but ‘very happy’ individuals
were signicantly more willing to sacrice than the ‘pretty
happy’ individuals, MD(SE) = .16(.07), p = .019 (Figure 1) – an
eect that remained signicant after controlling for age,
sex, income, and education (Table 2). There were no dier-
ences, however, between ‘very happy’ individuals and ‘not
too happy’ on how willing they were to sacrice. Only
education signicantly predicted greater willingness to
sacrice, β
income
= .11, p< .001.
Summary
In a nationally representative sample of Americans, we
obtained evidence that very happy individuals were
more, not less, likely to take environmental action – even
though people who were not happy worried more about
climate. This suggests that, even if happy people worry less,
they are more likely than unhappy people to take action in
their daily lives towards being environmentally conscious.
Notably, very happy individuals were also more willing to
engage in future sacrices than pretty happy individuals.
Although this study found no dierence in willingness to
sacrice between very happy and unhappy individuals, this
seems to be due to insucient power as fewer people
reported being ‘not happy’ (see Figure 1). This distinction
provides evidence against the argument that making peo-
ple happier will decrease their willingness to act on societal
issues. These eects were relatively robust to controlling for
a key possible confound – income – as well as other
demographics.
Discussion
Are happy people less likely to be involved in social,
political, and environmental action? Contrary to this
Pollyanna hypothesis, we found evidence across three
studies and various social issues that happiness predicts
more, not less, social action. Although these associa-
tions were statistically small, they generally revealed
that happier people tend to engage in more current
action in response to environmental threats, local social
issues, and global social issues that they care about. The
size of these associations remained robust after control-
ling for demographic factors in a nationally representa-
tive sample. Furthermore, we observed associations
between positive aect and action in demographically
homogenous samples (college students at an elite uni-
versity, where the students tended to be upper-middle-
class, largely Caucasian, and liberal in political attitudes)
making it unlikely that factors, such as income or edu-
cation, are the key drivers of the observed associations.
Thus, while causality cannot be inferred from correla-
tional data, we found no evidence for the Pollyanna
hypothesis: Happy people do not seem to be too self-
involved to participate in social action or strive to enact
change across local and global issues.
In contrast to happy people’s propensity for action
and involvement, unhappy people – though sometimes
concerned about important current issues – were no
more likely to act when given the opportunity to do so
in the context of our studies (e.g., signing up for email
newsletter to stay informed and engaged with an
important to them issue) or in their daily life (e.g.
recycling or conserving energy). Notably, however,
situationally-bound negative feelings about the specic
issue did predict more action.
Theoretical integration and implications
The overall pattern of observed relationships is consis-
tent with the aect-as-cognitive feedback model
(Huntsinger, Isbell, & Clore, 2014). According to this
perspective, positive aect serves as a cognitive ‘Go!’
signal, whereas negative aect serves as
a corresponding ‘Stop!’ signal. Thus, when happy peo-
ple feel driven by negative emotions to participate in
social, political, or environmental action, their brighter
Table 2. ANCOVA results comparing social action between very happy, pretty happy, and not too happy individuals in GSS
controlling for major demographics.
Concern Current Behavior Future Action
F t p F t p F t p
Omnibus Happiness 3.17 .042 2.93 .054 3.25 .039
Age 12.31 <.001 10.53 .001 1.22 .270
Sex 4.75 .030 3.89 .049 0.77 .381
Income 6.01 .014 0.58 .447 0.00 .969
Education 0.15 .702 19.44 < .001 13.55 < .001
Pairwise
Very Happy ←→ Pretty Happy −1.47 .305 2.41 .043 2.51 .032
Very Happy ←→ Not Happy −2.50 .033 0.96 .601 0.88 .653
Pretty Happy ←→ Not Happy −1.66 .220 −0.78 .713 −0.96 .602
Note. P-values in pairwise comparisons are Tukey corrected.
8K. KUSHLEV ET AL.

mood may catalyze such intentions to act into actual
behavior. In contrast, when unhappy people feel con-
cerned about an issue or driven to act, their somber
mood may impede this intention from turning into
tangible eort and action.
The pattern of ndings for negative aect is further
consistent with the feelings-as-information model
(Schwarz, 2012) by showing that high negative aect
predicts greater concern when thinking about various
issues. The feelings-as-information perspective, how-
ever, fails to explain why we observed the same – and
even somewhat larger – association between positive
aect and concern about issues. This apparent theore-
tical discrepancy may be due to a methodological dis-
crepancy: Whereas most of the research underlying the
feelings-as-information theory is based on temporary
moods induced in the lab (Schwarz, 2012), we exam-
ined people’s frequency of experiencing positive emo-
tions over an entire month. Thus, while watching
a funny video in the lab may put people in a giddy
mood, leading them to feel less worried about current
issues, being generally in a good mood across one’s
daily life may serve as an emotional resource: Feeling
good most of the time may allow people to feel, or
even actively cultivate feelings of, anger, indignation, or
worry about issues that they see as impacting their
community, society, and the world. Further research
and theory are needed to distinguish between the
cognitive eects of momentary positive moods versus
those of being happy.
Boundary conditions
We began this article by reviewing a growing literature
on the eects of negative emotions – from anger to
shame – on behavior aimed to engender personal,
social, or political change (Kashdan & Biswas-Diener,
2015). Do the observed negative-to-null associations
of being unhappy with social action contradict this
past research? We do not think so. Indeed, even though
general negative aect did not predict greater action
across most of our studies, negative aect experienced
in response to a specic issue did generally predict
more action and engagement. This pattern ts with
functional accounts of human emotion (Darwin, 1872�;
Ekman, 1992) by suggesting that negative emotions are
useful only in as much as they serve motivational pur-
poses in the context of specic events (but not when
being also elicited by benign stimuli). The contribution
of the present research is to distinguish between con-
text-bound negative emotions and feeling bad in
general.
It is important to note that the eects we observed
were statistically small. Of course, given the great num-
ber of demographic and psychological factors that
likely play a role in motivating social action, from
income to values or free time, we should hardly expect
any larger eects of positive aect – only one factor
among many. Still, it is also important to point out that
statistically small eects are not necessarily practically
insignicant. Thus, for example, even though Study 3
(GSS) produced some of the smallest correlation coe-
cients, we observed that going from ‘not too happy’ to
‘very happy’ was associated with up to 50% greater
engagement in behavior that is costly to oneself but
benecial to society (such as donating money to envir-
onmental causes). Ultimately, however, we note that
the size of the statistical eects is immaterial to the
primary theoretical question behind this research: To
examine whether, as increasingly suggested by popular
and academic writers, promoting personal happiness
may be counterproductive to societal and environmen-
tal stewardship. On the contrary, negative, rather than
positive, aect is associated with inaction in response to
local issues and global threats.
Future research
In the present research, we set out to systematically
examine the nature of the association of positive and
negative aect with action. Our research, therefore, in
no way can, or attempts to, address the issue of caus-
ality. While we did not explore causality, the positive
associations we observed do suggest that being happy
is not inherently bad for engaging in meaningful action.
Thus, our ndings do suggest that programs and initia-
tives aimed at increasing happiness are unlikely to
result in a world full of Pollyannas unmoved by impor-
tant current issues. Our research, however, in no way
suggests that interventions designed to raise happiness
would raise people’s engagement in social and political
activity. Future research needs to examine whether, as
suggested by Fredrickson’s Broaden and Build Model
(Fredrickson & Losada, 2005), the causal path may ow
both ways, whereby greater positive emotions lead to
more engagement in social and political action, which,
in turn, may foster greater happiness.
Coda
In contrast to the backlash against positive psychology
and positivity, Stephen Pinker (2011) has argued that far
from suering from unrealistic optimism, people across
the globe tend to be unrealistic pessimists, viewing the
THE JOURNAL OF POSITIVE PSYCHOLOGY 9

world through an increasingly negative and politicized
narrative despite an immense progress of human civiliza-
tion and the unprecedented quality of life of most modern
humans. In the present work, we have shown that we
need not worry that such calls to be more positive, hap-
pier, and optimistic could jeopardize this progress by turn-
ing people into inactive participants in the challenges
faced by communities, nations, and humanity.
Note
1. We also measured other components of well-being,
including life satisfaction, which are not the focus of
the present report. Because the results of life satisfac-
tion mirrored the results of PA, we decided to focus this
report on aect.
Disclosure statement
No potential conict of interest was reported by the authors.�
ORCID
Kostadin Kushlev http://orcid.org/0000-0003-1025-3258
References
Adler, S. E. (2018). The power of negative thinking. https://
www.theatlantic.com/magazine/archive/2018/01/the-
power-of-negativity/546560/�
Boehm, J. K., & Lyubomirsky, S. (2008). Does happiness pro-
mote career success? Journal of Career Assessment,16(1),
101–116.
Boehm, J. K., Vie, L. L., & Kubzansky, L. D. (2012). The promise
of well-being interventions for improving health risk
behaviors. Current Cardiovascular Risk Reports,6(6),
511–519.
Bohart, A., & Greening, T. (2001). The American Psychologist.
Humanistic Psychology and Positive Psychology,56(1),
81–82–90.
Chang, E. C. (2008). Self-criticism and self-enhancement: Theory,
research, and clinical implications. Washington, DC:
American Psychological Association.�Darwin, C. (1872). The
expression of emotions in man and animals. Chicago:
University of Chicago Press.
Diener, E., & Chan, M. Y. (2011). Happy people live longer:
Subjective well-being contributes to health and longevity.
Applied Psychology: Health and Well-Being,3(1), 1–43.
Diener, E., Wirtz, D., Tov, W., Kim-Prieto, C., Choi, D. W.,
Oishi, S., & Biswas-Diener, R. (2010). New well-being mea-
sures: Short scales to assess ourishing and positive and
negative feelings. Social Indicators Research,97(2), 143–156.
Ein-Dor, T., Mikulincer, M., & Shaver, P. R. (2011). Eective
reaction to danger: Attachment insecurities predict beha-
vioral reactions to an experimentally induced threat above
and beyond general personality traits. Social Psychological
and Personality Science,2(5), 467–473.
Ekman, P. (1992). An Argument for Basic Emotions. Cognition
and Emotion,6(3–4), 169–200.
Ferrer, R. A., Klein, W. M. P., Zajac, L. E., Sutton-Tyrrell, K.,
Muldoon, M. F., & Kamarck, T. W. (2012). Unrealistic opti-
mism is associated with subclinical atherosclerosis. Health
Psychology,31(6), 815–820.
Forgas, J. P. (2013). Don’t worry, be sad! On the cognitive,
motivational, and interpersonal benets of negative mood.
Current Directions in Psychological Science,22(3), 225–232.
Forgas, J. P. (2014). On the downside of feeling good. In
J. Gruber & J. T. Moskowitz (Eds.), Positive emotion:
Integrating the light sides and dark sides (pp. 301–322).
New York, NY: Oxford University Press.
Fredrickson, B. L., & Losada, M. F. (2005). Positive aect and
the complex dynamics of human ourishing. American
Psychologist,60(7), 678–686.
Grant, N., Wardle, J., & Steptoe, A. (2009). The relationship
between life satisfaction and health behavior: A
cross-cultural analysis of young adults. International
Journal of Behavioral Medicine,16(3), 259–268.
Huntsinger, J. R., Isbell, L. M., & Clore, G. L. (2014). The aective
control of thought: Malleable, not xed. Psychological
Review,121(4), 600–618.
Iyer, A., Schmader, T., & Lickel, B. (2007). Why individuals
protest the perceived transgressions of their country: The
role of anger, shame, and guilt. Personality and Social
Psychology Bulletin,33(4), 572–587.
Kashdan, T., & Biswas-Diener, R. (2015). The upside of your dark
side: Why being your whole self—Not just your “good” self—
Drives success and fulllment. New York, NY: Plume.�
Keltner, D., & Lerner, J. S. (2010). Emotion. In S. T. Fiske,
D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychol-
ogy (Vol. 1, 5th ed., pp. 317–352). Hoboken NJ: Wiley.
Kim, E. S., Kubzansky, L. D., & Smith, J. (2015). Life satisfaction
and use of preventive health care services. Health Psychol,
34(7), 779–782.
Knight, B. (2014). The power of negative thinking: An unconven-
tional approach to achieving positive results. Boston, MA:
New Harvest, Houghton Miin Harcourt.
LaBier, D. (2017). Can embracing negative emotions increase
your well-being? Psychology Today.https://www.psycholo
gytoday.com/us/blog/the-new-resilience/201709/can-
embracing-negative-emotions-increase-your-well-being
Lang, F. R., Weiss, D., Gerstorf, D., & Wagner, G. G. (2013).
Forecasting life satisfaction across adulthood: Benets of
seeing a dark future? Psychology and Aging,28(1), 249–261.
�
Lazarus, R. S. (2003). Does the positive psychology movement
have legs? Psychological Inquiry,14(2), 93–109.
Lyubomirsky, S., King, L., & Diener, E. (2005). The benets of
frequent positive aect: Does happiness lead to success?
Psychological Bulletin,131(6), 803–855.
Oettingen, G. (2000). Expectancy eects on behavior depend
on self-regulatory thought. Social Cognition,18(2), 101–129.
Oishi, S., Diener, E., & Lucas, R. E. (2007). The optimum level of
well-being: Can people be too happy? Perspectives on
Psychological Science,2(4), 346–360.
Pinker, S. (2011). The better angels of our nature. New York, NY:
Viking.
Pressman, S. D., & Cohen, S. (2005). Does positive aect inu-
ence health? Psychological Bulletin,131(6), 925–971.
10 K. KUSHLEV ET AL.

Roseman, I. J., Wiest, C., & Swartz, T. S. (1994).
Phenomenology, behaviors, and goals dierentiate discrete
emotions. Journal of Personality and Social Psychology,67
(2), 206–221.
�Schwarz, N. (2012). Feelings-as-information theory. In P. A. M.
Van Lange, A. W. Kruglanski, & E. T. Higgins
(Eds.), Handbook of theories of social psychology (pp.
289-308). Thousand Oaks, CA, : Sage Publications Ltd.
http://dx.doi.org/10.4135/9781446249215.n15.�
Tangney, J. P., Stuewig, J., & Mashek, D. J. (2007). Moral emo-
tions and moral behavior. Annual Review of Psychology,58,
345–372.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development
and validation of brief measures of positive and negative
aect: The PANAS scales. Journal of Personality and Social
Psychology,54(6), 1063–1070.
Weinstein, N. D., & Lyon, J. E. (1999). Mindset, optimistic bias
about personal risk and health-protective behaviour. British
Journal of Health Psychology,4(4), 289–300.
.�
Weitz-Shapiro, R., & Winters, M. S. (2011). The link between
voting and life satisfaction in Latin America. Latin American
Politics and Society,53(4), 101–126.
THE JOURNAL OF POSITIVE PSYCHOLOGY 11