DP 14447

DISCUSSION PAPER SERIES
IZA DP No. 14447
A Local Community Course That Raises

Wellbeing and Pro-sociality: Evidence
from a Randomised Controlled Trial
Christian Krekel
Jan-Emmanuel De Neve
Daisy Fancourt
Richard Layard
JUNE 2021
DISCUSSION PAPER SERIES
IZA DP No. 14447

Christian Krekel Daisy Fancourt

LSE and IZA University College London
Jan-Emmanuel De Neve Richard Layard
University of Oxford LSE and IZA
JUNE 2021
Any opinions expressed in this paper are those of the author(s) and not those of IZA. Research published in this series may
include views on policy, but IZA takes no institutional policy positions. The IZA research network is committed to the IZA
Guiding Principles of Research Integrity.
The IZA Institute of Labor Economics is an independent economic research institute that conducts research in labor economics
and offers evidence-based policy advice on labor market issues. Supported by the Deutsche Post Foundation, IZA runs the
world’s largest network of economists, whose research aims to provide answers to the global labor market challenges of our
time. Our key objective is to build bridges between academic research, policymakers and society.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper
should account for its provisional character. A revised version may be available directly from the author.
ISSN: 2365-9793
IZA – Institute of Labor Economics

Schaumburg-Lippe-Straße 5–9 Phone: +49-228-3894-0
53113 Bonn, Germany Email: publications@iza.org www.iza.org
IZA DP No. 14447 JUNE 2021
ABSTRACT
Despite a wealth of research on its correlates, relatively little is known about how to
effectively raise wellbeing in local communities by means of intervention. Can we teach
people to live happier lives, cost-effectively and at scale? We conducted a randomised
controlled trial of a scalable social-psychological intervention rooted in self-determination
theory and aimed at raising the wellbeing and pro-sociality of the general adult population.
The manualised course (“Exploring What Matters”) is run by non-expert volunteers
(laypeople) in their local communities and to date has been conducted in more than
26 countries around the world. We found that it has strong, positive causal effects on
participants’ subjective wellbeing and pro-sociality (compassion and social trust) while
lowering measures of mental ill health. The impacts of the course are sustained for at least
two months post-treatment. We compare treatment to other wellbeing interventions and
discuss limitations and implications for intervention design, as well as implications for the
use of wellbeing as an outcome for public policy more generally.
JEL Classification: C93, I12, I31

Keywords: wellbeing, pro-social behaviour, communities, intervention, RCT
Corresponding author:
Christian Krekel
London School of Economics
Department of Psychological and Behavioural Science
Centre for Economic Performance (CEP)
Houghton Street
London WC2A 2AE
United Kingdom
E-mail: c.krekel@lse.ac.uk
COMMUNITY COURSE FOR WELLBEING AND PRO-SOCIALITY 2
Introduction
For decades, enormous academic effort has been put into exploring the causes and conse-
quences of wellbeing (Diener et al., 1999; Layard et al., 2014). Health (especially mental
health), being partnered, and social relationships account for more than three quarters of the
explained variance in adult people's life satisfaction (Clark et al., 2018). At the same time,
there is growing evidence showing that wellbeing is a significant predictor of important life
and economic outcomes, including health and longevity (Danner et al., 2001; Steptoe and
Wardle, 2011; Graham and Pinto, 2019), productivity and income (De Neve and Oswald,
2012; Oswald et al., 2015; Bellet et al., 2020), voting (Liberini et al., 2017), and even compli-
ance with lockdown measures during Covid-19 (Krekel et al., 2020).
Yet, we know little about how to effectively improve the wellbeing of the general
adult population. Can we teach people to live happier lives? Can we do this by means of inter-
vention, cost-effectively and at scale? Are impacts sustained over time? Answering these
questions has profound implications: if wellbeing is not fixed and can be taught, it can be used
as a meaningful indicator to measure societal progress, and help direct public policy attention
towards areas that are malleable and where there is room for improvement.
The answers to these questions, however, are not ex-ante clear. A prominent view ar-
gues that there exists a set point of wellbeing around which individuals fluctuate (Brickman
and Campbell, 1971). According to this view, individuals largely adapt to various changes in
life circumstances, driven by withdrawal of attention to these changes, so that their wellbeing
remains largely unchanged over time (Frederick and Loewenstein, 1999; Kahneman, 2000).
Hedonic adaptation has been used to explain phenomena such as why life satisfaction has
been stagnant in many developed countries over the past decades while economic living
standards have increased substantially (Easterlin, 1974, 2010). There is now an established
body of evidence on hedonic adaptation to various positive or negative changes in life circum-
stances, including changes in marital status (Lucas, 2005; Lucas and Clark, 2006; Oswald and
Gardner, 2006; Stutzer and Frey, 2006), disability (Menzel et al., 2002; Oswald and Powdt-
havee, 2008), or income (Di Tella et al., 2010; Kuhn et al., 2011). According to this view
then, wellbeing is less malleable and significant increases in average population wellbeing
may be limited in societies with already high economic living standards.
Another point of view, in line with expectancy-value theory in psychology (Battle,
1965), suggests that familiarising people with evidence on what could make them happier
may lead to an update in their beliefs, which, in turn, may lead to a change in their behaviour.
Expectancies refer to the subjective probabilities of becoming happier which are attached to
certain behaviours, whereas values refer to the magnitudes of happiness changes resulting
from these behaviours. To the extent that this change in their behaviour may improve people's
wellbeing and thereby reinforce their beliefs, people may uphold that behaviour, leading to
permanent (as opposed to temporary) wellbeing change. This mechanism may be especially
effective when it comes to behaviours in life domains which are important for wellbeing and,
at the same time, are less prone to hedonic adaptation, such as time spent on social relation-
ships (Powdthavee, 2008), experiences (Carter and Gilovich, 2010), or pro-social action
(Dunn et al., 2008; Aknin et al., 2013; Drouvelis and Grosskopf, 2016). According to this in-
terventionist view, wellbeing is malleable and significant increases in average population
wellbeing may be possible, even in economically affluent societies.
Interventions that aim to improve wellbeing directly have typically been narrow in fo-
cus, looking at specific, often clinical target groups or at-risk populations (as opposed to
healthy adults in the general population), often including people suffering from depression
and anxiety (see Taylor et al. (2017), for example) or bodily pain (see Hausman et al. (2014),
for example).1 A notable exception is Heintzelman et al. (2019): the authors evaluated the im-
pact of ENHANCE, a 12-week wellbeing course targeted at the general adult population in
1
See Sin and Lyubomirsky (2009) and Bolier et al. (2013) for meta-analyses.
their local communities which has been trialled in hybrid (i.e. ten sessions online and two ses-
sions offline) and face-to-face (i.e. twelve sessions offline) delivery. When delivered face-to-
face, it is led by graduate-level trained clinicians. Similar to the intervention presented in this
paper, it focuses primarily on positive habits, skills, and attitudes. During the course, a new
skill is introduced every week, participants practice that skill, and then write about their expe-
riences. The authors found that it had strong, positive causal effects on participants' wellbeing
up to six months after the main intervention has ended and up to three months after an ex-
tended following-up period.
We studied the impact of a similar course – "Exploring What Matters" – which is a lo-
cal community intervention aimed at raising the wellbeing and pro-sociality of the general
adult population. Besides contents, it differs from existing interventions in at least two critical
aspects: first, the course is manualised and led by non-expert volunteers (laypeople) rather
than trained clinicians, making it highly cost-effective. Second, due to its cost-effectiveness, it
is highly scalable and can be delivered face-to-face in the local communities of course leaders
and participants. Cost-effectiveness and scalability have important implications for the feasi-
bility of social prescribing in health economics, i.e. the referral by GPs to non-medical com-
munity interventions to address the wider determinants of health and to help patients improve
health-related behaviours (see NHS Long Term Plan (2019), for example). As of August
2020, 431 courses have been completed, with a total of 5,621 participants, yielding an average
course size of 15 (13 course participants plus two volunteers leading the course). Most
courses have been conducted in the UK (343), with a further 88 courses run in 25 countries.
"Exploring What Matters" is run by Action for Happiness, a registered charity in England,
which was launched in 2011. Its patron is the Dalai Lama, who helped to launch the course in
London in 2015.
Using a randomised controlled trial, we studied the impacts of six of these courses
which took place in London between August 2016 and December 2017: two during autumn
2016, two during spring 2017, and two during autumn 2017. In what follows, we first de-
scribe the intervention, derive hypotheses on wellbeing change, and illustrate the study de-
sign, before turning to our findings on self-reported outcomes and biomarkers. We then pre-
sent the results of a replication exercise using before-after data from the universe of courses
conducted to date. Finally, we calculate the cost-effectiveness of the course in raising wellbe-
ing, compare it to other interventions in the literature, and discuss shortcomings, implications,
and avenues for future research in the field, as well as implications for the use of wellbeing as
an outcome for public policy more generally.
The Intervention
The "Exploring What Matters" course brings together participants in face-to-face groups to
discuss what matters for a happy, meaningful, and virtuous life. Participants span a wide
range of ages and socio-economic backgrounds but can be broadly classified, as per their self-
reports, into two categories: people who are unhappy and looking for ways to improve their
lives; and people who are interested in wellbeing more generally and want to learn more, or
want to share these ideas with others.
The intervention is manualised: each course is led by two volunteers as facilitators on
an unpaid basis.2 Recruitment of course leaders follows a documented, standardised process:
each candidate completes a Leader Registration process sharing their motivation and experi-
ence and is given instructions on what is required. Once potential course leaders have a co-
leader, venue, and dates in mind, they complete a Course Application process. Action for
Happiness reviews this application and, if certain criteria are met, arranges a call to discuss
2
Although the intervention is manualised, some degree of adaptation is possible. For example, course
leaders may choose the most appropriate venue or allow for more group discussion time. However, they are en-
couraged to stick closely to the course guide.
next steps.3 Once a course is approved, course leaders receive guidance and structured re-
sources to facilitate course delivery. Supplementary Materials II includes a link to the com-
plete documentation of the recruitment process of course leaders.
Participants sign up online, and when doing so, are asked to make a donation; dona-
tions aim to cover the implementation costs of the course (implementation costs are about £90
or $113 per participant, including variable costs for course materials as well as allocated fixed
costs).4 Donations are voluntary and participants can take part without donating. The function
of donations is to make the course scalable and accessible to people regardless of their finan-
cial situation. Besides that, they aim at raising course attendance, by exploiting the notion of
sunk costs.5 The course consists of eight consecutive weekly sessions lasting between two and
2.5 hours each. Each of these sessions builds on a thematic question, for example, what mat-
ters in life, how to find meaning at work, or how to build happier communities. Each of these
questions is discussed against the background of scientific evidence on subjective wellbeing,
mental health, and pro-sociality as well as motivation and group learning.
Courses are advertised both online and offline in local communities, and potential par-
ticipants must register online. Online advertising is done via emails to people who have previ-
ously registered with Action for Happiness and live nearby and to new people via targeted lo-
cal Facebook advertising. Offline advertising is done via local course leaders using word-of-
mouth and, to a lesser extent, local promotion (for example, through notice boards or local
press).
3
Course leaders have a similar demographic profile as course participants, with a slightly higher aver-
age age. 58% are female. 58% are between 31 and 50 years old, 25% between 18 and 30, and 17% between 51
and 70. They tend to have higher than average levels of life satisfaction and social trust (both about 7.9 on zero-
to-ten scales).
4
Converted using an exchange rate of 1:1.25 as of July 16, 2020.
5
Unfortunately, we did not have data on the donation amount per participant, and hence could not study
heterogeneity of course outcomes depending on donations.
Hypotheses. Course design and delivery are rooted in psychological self-determina-
tion theory (Deci and Ryan, 1985), which states that autonomy, relatedness, and competence
are fundamental human needs that enable people to achieve wellbeing. The course aims at
building (i) autonomy by enabling participants to discover for themselves what matters for
their lives, using a weekly mindfulness exercise, gratitude exercise, and personal reflection,
supported by a "Did You Know?" section that introduces scientific evidence on that week's
theme; (ii) relatedness by facilitating interpersonal connections and social trust, within the
gathering of people in their local communities; and (iii) competence by enabling participants
to experience for themselves how behavioural changes to daily routines can make differences
to their and other people's wellbeing, using goal-setting and social commitment tools to help
translate motivation into action. Supplementary Materials II includes links to the complete
course materials of both course participants and course leaders.
There is an established evidence base linking psychological self-determination theory
to wellbeing (Ryan and Deci, 2000), across life domains and different cultural contexts (Mi-
lyavskaya and Koester, 2011; Church et al., 2012), including its constituent elements (Brown
et al., 2003; Chirkov et al., 2003; La Guardia et al., 2000). Likewise, there is evidence from
systematic reviews and meta-analyses linking certain elements of the course curriculum, in
particular mindfulness, meditation, and related self-regulation strategies, to positive outcomes
in non-clinical populations, including wellbeing, depression, and anxiety, with medium to
strong effect sizes (see Sedlmeier et al. (2012), Gu et al. (2015), or Querstret et al. (2020), for
example).
We therefore hypothesise that, first, the course has positive impacts on wellbeing. Sec-
ond, we hypothesise that – to the extent that it fosters interpersonal connections between
strangers and encourages pro-social action-taking – the course has positive impacts on pro-
social attitudes. Third, we hypothesise that – to the extent that it changes beliefs about
behaviours in life domains that are important for wellbeing and that are less prone to hedonic
adaptation – the course may have sustained impacts.
Methods
We conducted a randomised controlled trial which focused on six courses that took place in
London between August 2016 and December 2017, including a total of 146 participants.
These were informed about the study, both during online registration and on site, and written
consent was taken.6 Following power calculations based on the historical number of course
participants (about 13 per course), this sample size was determined before any data collection
and analysis.7
Course participants were self-selected. To study the extent to which they differed from
the general adult population, we compared our estimation sample, pre-treatment, with a sam-
ple from the nationally representative UK Household Longitudinal Survey ("Understanding
Society"), restricted to London and to the same age span as our participants. We found that
there were little, quantitatively relevant differences in the age distribution between course par-
ticipants and the general population. Participants were, however, significantly more likely to
be female in our sample (83% vs. 45%). Moreover, they were significantly less likely to be
married (20% vs. 53%) and more likely to be in a domestic partnership (25% vs. less than one
percent). This difference, however, is likely to be an artefact arising from survey design: Un-
derstanding Society does not ask about a "domestic" (as our survey did) but about a "civil"
partnership. When it comes to income, we found again little, quantitatively relevant differ-
ences, except for the highest income category: our sample included significantly less
6
This study passed the Internal Review Board of the Research Ethics Division at the London School of
Economics (Reference: 00507).
7
A power of 0.8, alpha of 0.05 two-tailed, and an assumed effect size of 0.5 yielded at least N=128 in-
dividuals, with 64 per experimental group.
individuals earning £75,000 ($94,000) or more and was somewhat more skewed towards
lower incomes. Finally, participants reported, on average, a lower level of life satisfaction (by
about 47% of a standard deviation), pre-treatment, than the general population.8
Randomised Controlled Trial. To account for self-selection of participants into the
course, we employed a waitlist randomisation protocol: after registering for the course online,
participants (who reported that they were able to attend the course on either one of two sets of
pre-specified upcoming dates, two months apart) were randomly allocated to one of the two
sets, unaware of how these related to treatment and control group. Participants in the earlier
set of dates were in the treatment group, those in the later set in the waitlisted control group.
They were then invited to arrive on the same date to have their first data collected. The event
started with a brief introductory session which explained to participants that they were re-
quired to fill in surveys and provide saliva samples. This was when participants read the pro-
ject information sheet and signed written consent forms. After written consent had been ob-
tained, the data were collected. After data collection had finished, the brief introductory ses-
sion was over and participants in the treatment group started their course immediately. Partici-
pants in the control group would start their course eight weeks later, after the treatment group
would have finished, and left the premises. Treatment and control group were kept separate:
neither group knew anything about the other, and the two groups did not meet on that day.
Note that the choice of the appropriate control group is not trivial: as there exists no
natural, credible counterfactual that could lend itself as a business-as-usual scenario in our in-
tervention context, choosing a waitlisted control group comprised of those who initially se-
lected into the intervention seems most appropriate for adhering as closely as possible to evi-
dence-based practice. Note that our control group does not include a placebo: arguably, a pla-
cebo could help to better isolate and identify the active ingredients of the intervention. At the
8
See Supplementary Materials Table 1a for this analysis.
same time, however, it raises the question of what precisely the (neutral) placebo can be,
whether one control group with one placebo is actually enough, and whether or not elements
like socialising are active parts of the intervention package and should thus be accounted for
as such. We will return to the issue of choosing the appropriate control group in more detail
later on in the discussion section.
Data Collection. Data were collected at three points in time: at t=0, right before the
course started; at t=1, right before it ended, which was eight weeks after t=0; and at t=2, eight
weeks after t=1. At each point in time, data were collected at the same hour of day (circa 6pm
in the evening). Figure 1 illustrates our randomised controlled trial and data collection pro-
cess.
Figure 1
Randomised Controlled Trial and Data Collection
Sources: Own illustration.

Our estimation sample (exploiting data points at t=0 and t=1) consisted of 146 re-
spondents (279 observations), of which 73 were in the treatment (136 observations) and 73
(143 observations) were in the control group. As can be seen, in our estimation sample, we
have an attrition rate of about 5%.9 We will test the sensitivity of our results regarding attri-
tion later in our robustness section. To look at treatment effect persistence, we exploited data
points at t=2 in an extended sample. As all respondents had been treated at t=2, results are ex-
ploratory.
Importantly, data at t=0 and t=1 were collected right before the start of the first and the
last session, respectively, at the back of the meeting room. Collecting data before the start of
the respective session reduced measurement error which may have resulted from participants'
euphoria of having started or finished the course being mixed up with actual outcomes. Note
that, during data collection at t=0 and t=1, the atmosphere was deliberately kept neutral, and
participants were asked to complete surveys and give biomarker samples before they had a
chance to meet other participants in the main room. To be consistent, the same protocol re-
garding neutrality of atmosphere that applied to data collection at t=0 and t=1 also applied to
data collection at t=2. Attending data collection at t=2 had been communicated as mandatory
beforehand. To avoid creating emotional arousal about attending this additional session, par-
ticipants did not know what content and format it involved. Finally, neither course partici-
pants nor volunteers leading the course knew whether they were in the treatment or control
group during data collection at t=0. Participants' group allocation was announced only after
data collection at t=0 had finished.
Outcomes. We collected data on two categories of outcomes: self-reported outcomes
came from survey data, which included items on subjective wellbeing, mental health, and pro-
sociality. Biomarkers were collected through saliva samples, which included cortisol – a
9
That is, ((279/(146*2))-1)*(-1)=~0.05.
steroid hormone responsive to stress – and a range of cytokines – immune proteins involved
in inflammatory response. Activation of the inflammatory response system has been shown to
be bidirectionally associated with mental ill health and depressive symptoms (Dowlati et al.,
2010; Miller and Raison, 2016). Supplementary Materials III contain the project information
sheet, written consent form, and the survey instruments used in the study, including surveys at
t=0, t=1, and t=2.
Items on subjective wellbeing covered evaluative (life satisfaction), experiential (hap-
piness and anxiousness), and eudemonic (worthwhileness) dimensions. They were measured
on eleven-point single-item Likert scales whereby zero denoted the lowest possible level and
ten the highest. Items on mental health covered frequently used screening measures to detect
depression (the three-point nine-item Patient-Health Questionnaire, PHQ-9) and anxiety (the
three-point seven-item Generalised-Anxiety-Disorder Questionnaire, GAD-7). PHQ-9 scores
from zero to four imply minimal, from five to nine mild, from ten to fourteen medium, and
from fifteen to 27 strong depression symptomatology. GAD-7 scores have a similar interpre-
tation but are cut off at 21. Respondents in our sample could thus be characterised as, on aver-
age, mildly depressed (M=6.4, SD=4.5) and anxious (M=6.1, SD=4.6). Distributions were,
however, highly skewed: in case of depression, for example, we found that 24 out of 133 re-
spondents for whom we had data at t=1 (about 18%) showed medium or strong depressive
symptomatology. When these were omitted, the remaining respondents could be characterised
as only minimally depressed (M=4.4, SD=2.7), not much different from PHQ-9 scores typi-
cally found at the general adult population level, which range from M=3.0, SD=4.3 for 30 to
39 year olds to M=3.7, SD=5.1 for 50 to 59 year olds in the US, for example (Tomitaka et al.,
2018). Items on pro-sociality included the Santa Clara Brief Compassion Scale – a composite
score running from five to 35 which measures pro-sociality by asking respondents about their
readiness to help others – and eleven-point single-item Likert scales on social trust and
gratitude. We standardised self-reported outcomes to have mean zero and standard deviation
one, using the course-set-specific control group mean and standard deviation.
Biomarkers included, besides cortisol, pro-inflammatory cytokines IL-1 and IL-6,
anti-inflammatory cytokine IL-10, interferon IFN- , and chemokine IL-8. These markers have
been shown to be responsive to both short-term and long-term psychosocial interventions
(Fancourt et al., 2016). They were collected by means of a saliva sample right after the sur-
veys with self-reported outcomes had been completed. We applied passive drool method of
sample collection using low protein-bind collection cryovials. Samples were analysed – three
times independently at the Institute for Interdisciplinary Salivary Bioscience Research at the
University of California at Irvine – using multiplex immunoassays. Cortisol was measured in
µg/dL, cytokines in pg/mL. We took means across the three analyses run for each biomarker,
removed outliers, and log-transformed and standardised the data.
Controls. We collected survey data on socio-demographic characteristics of respond-
ents, including age, gender, marital status, education, employment, income, religion, religious
practice, preference for meeting new people and making new friends, health (including preg-
nancy), and health-related behaviours (including smoking and medication usage), to control
for potential differences between treatment and control group over time. All controls were
measured pre-treatment. Table 1b in the Supplementary Materials shows variable definitions
and descriptive statistics, Table 1c balancing properties between treatment and control group:
there was little evidence for significant mean differences in outcomes and controls between
groups prior to course start. Similarly, Table 4 in the Supplementary Materials shows that
there was little evidence for significant differences for the control group between t=0 and t=1,
pointing towards the absence of time trends or waitlist effects. There were no known con-
founding events during the study period.
Descriptive Evidence. Before turning to our empirical estimation, we first look at se-
lected descriptive evidence on subjective wellbeing, mental health, and pro-sociality. Figure 2
plots the raw means of four of our self-reported outcomes – life satisfaction, mental health
(PHQ-9 for depression and GAD-7 for anxiety), and social trust – during the observation pe-
riod.10
Figure 2
Average Scores of Groups at Different Points in Time
Life Satisfaction PHQ-9 (Depression)

8,0 8,0
7,5 7,0
7,0 6,0
6,5 5,0
6,0 4,0
5,5 3,0
5,0 2,0
t=0 t=1 t=2 t=0 t=1 t=2
Treatment Group Control Group Treatment Group Control Group
GAD-7 (Anxiety) Social Trust

8,0 8,0
7,0 7,5
6,0 7,0
5,0 6,5
4,0 6,0
3,0 5,5
2,0 5,0
t=0 t=1 t=2 t=0 t=1 t=2
Treatment Group Control Group Treatment Group Control Group
Notes: A waitlist randomisation design was applied: between t=0 and t=1, the treatment group received treat-
ment; between t=1 and t=2, the control group received treatment. Scores are in natural units. Life satisfaction and
social trust were measured on scales from zero to ten, PHQ-9 for depression on a scale from zero to 27, and
GAD-7 for anxiety on a scale from zero to 21. N=383 (146 at t=0, 133 at t=1, and 104 at t=2). Confidence inter-
vals are 95%.
Sources: Own data collection, own calculations.
We make three observations: first, between points t=0 and t=1, the course improved the scores
of the treatment group, in line with our first and second hypotheses, whereas those of the
10
Figures for other self-reported outcomes are available upon request.
control group remained constant. Second, between points t=1 and t=2, the course improved
the scores of the control group (which received treated during that period) in a similar fashion,
whereas those of the treatment group were sustained or even continued to improve, in line
with our third hypothesis.
Empirical Model. We now turn to our empirical model. Our baseline model is a dif-
ference-in-differences specification that compared the evolution of course outcomes between
groups over time:11
yit = 0 + 1Treatmenti*Postt + 2Treatmenti + 3Postt + 4'Xit + μs + εit with t = {0, 1}
(1)
where yit is the outcome of respondent i at time t; Treatmenti is a dummy equal to one if the
respondent belonged to the treatment group, and zero else; Postt is a dummy equal to one at
t=1, and zero else; Xit is a vector of controls; and μs is a course-set-specific fixed effect. In
what follows, we present coefficients obtained from estimating Equation 1 without controlling
for Xit, and relegate those obtained from estimating the equation with the vector of controls to
the Supplementary Materials. If randomisation was successful and treatment was exogenous,
controlling for Xit should not make any difference, and this is precisely what we will show.
Our model was estimated using OLS, with robust standard errors clustered at the par-
ticipant level. 1 is the causal effect (average treatment effect on the treated) of course partici-
pation. Note that our model could not exploit data points at t=2 because there was no credible
control group anymore.
11
Alternatively, one could regress the post-treatment on the pre-treatment outcome and a treatment
dummy (which enforces a balanced panel). Results were qualitatively the same.
Taken together, we tested fifteen hypotheses in our main analysis (i.e. four outcomes
related to subjective wellbeing, two outcomes related to mental health, three outcomes related
to pro-sociality, plus six biomarkers). To account for multiple hypotheses testing, we used the
stepdown multiple testing procedure suggested by Romano and Wolf (2005a, 2005b), with the
four-step algorithm outlined in Romano and Wolf (2016). In essence, the algorithm constructs
a null distribution for each of our fifteen hypotheses tests based on a set of null resampling
test statistics (in our case, using a bootstrap with 100 repetitions and cluster-robust standard
errors at the participant level in both the original regression and during the resampling proce-
dure). We find that our stepdown adjusted P values (corresponding to the significance of a hy-
pothesis test where fifteen tests were implemented) continue to indicate significance at con-
ventional levels for all our coefficient estimates (where our original P values indicated signifi-
cance).12
Results
Impacts on Subjective Wellbeing, Mental Health, and Pro-Sociality. Figure 3 plots the co-
efficient estimates of our self-reported outcomes. We again confirmed our first and second
hypotheses on positive impacts on wellbeing and pro-social attitudes.
Figure 3
Impacts on Self-Reported Outcomes: Subjective Wellbeing, Mental Health, and Pro-Sociality
12
See Supplementary Materials Tables 5a and 5b for these results.
Notes: Outcomes have been standardised prior to running regressions (i.e. transformed to z-scores with mean of
zero and standard deviation of one, using the control group mean and standard deviation). See Supplementary
Materials Table 2a for the corresponding regression table with controls. Robust standard errors were clustered at
the participant level. N=279 (146 respondents, of which 73 were in treatment and 73 in control). Confidence
bands are 95%.
In terms of subjective wellbeing, the course increased life satisfaction by about 64% of a
standard deviation, happiness by about 63%, and worthwhileness by about 56%. Anxiousness,
on the contrary, was decreased by about 42%. Impacts were large: for life satisfaction, for ex-
ample, the effect size corresponds to an increase of about one point on a zero-to-ten scale. Im-
pacts were significant at the 5% level.
In terms of mental health, the course decreased both PHQ-9 and GAD-7 scores, re-
spectively, by about 54% and 45% of a standard deviation (impacts did not significantly differ
from each other). Impacts were again large: participants, prior to taking the course, reported
mean PHQ-9 and GAD-7 scores of about 6.7 and 6.1, respectively, which corresponds to a
clinical symptomatology of mild depression and anxiety. The course improved scores to, on
average, 4.3 points for PHQ-9 and 3.7 for GAD-7, which corresponds to minimal depression
and anxiety. Impacts were again significant at the 5% level.
Although strong, impacts on mental health were clearly weaker than those found in tri-
als based on cognitive behavioural therapy. For example, the Improving Access to Psycholog-
ical Therapies scheme in the UK has been found to reduce PHQ-9 and GAD-7 scores, on av-
erage, by about eight and seven (Clark et al., 2009). The Cognitive Behavioural Therapy as an
Adjunct to Pharmacotherapy trial has been found to reduce PHQ-9 and GAD-7 scores, on av-
erage, by about 7.1 and 4.7 (Wiles et al., 2016). However, these trials were targeted specifi-
cally at individuals who suffer from depression and anxiety, rather than the general adult pop-
ulation.
In terms of pro-sociality, we found that the course significantly increased both com-
passion and social trust at the 5% level, respectively, by about 42% and 56% of a standard de-
viation (about 0.6 and 1.1 points). The impact on gratitude, however, was lower and only mar-
ginally significant.
Next, we ran a series of regressions to look into the importance of social context, po-
tential mechanisms behind our average treatment effects, and heterogeneous effects. To do so,
we first re-estimated our baseline model with controls (Supplementary Materials Table 2a),
and then selectively exploited these controls in these subsequent analyses. Note that including
controls has little impact on our identified effects (compare Figure 3 with Supplementary Ma-
terials Table 2a), which suggests that randomisation was successful and treatment was exoge-
nous.
To study the importance of social context, we note that whether or not we control for
social context, measured as participants' preference for socialising, has made little difference
to our findings.13 Next, we ran two additional regressions. First, we re-estimated our model
without controlling for participants' preference for socialising but controlling for all other co-
variates: coefficient estimates were slightly attenuated yet continued to be strong, suggesting
that socialising may play a role but only partially explains impacts. We then split our sample
by the mean pre-treatment value of this variable: again, we did not find that impacts were sys-
tematically stronger for respondents who had a higher preference for socialising, pre-treat-
ment, and vice versa. Thus, it does not seem that participants who had a higher preference for
socialising benefited more from the course than others, or the other way around.14
To explore potential mechanisms, we collected data on two categories of additional
outcomes: information and behaviour. The former included measures that relate to knowledge
of what contributes to one's own and other people's wellbeing. The latter included measures
that relate to frequencies of behaviours in various social domains, including the private
sphere, close relationships, and other people.15 Items on information and behaviour also
served as manipulation checks, as the course explicitly aims at changing both.
Re-estimating our baseline model with standardised measures of information as out-
comes, we indeed found that participants reported to feel more knowledgeable of what con-
tributes to a happy and meaningful life, to know more what matters to them personally, and to
feel more able to do things to improve their own, and to a somewhat lesser extent, the wellbe-
ing of other people. When it comes to standardised measures of behaviour, the course in-
creased the frequency in which participants reported to practice mindfulness or meditation, to
treat themselves in a kind way, to connect with other people, and to do something kind or
helpful for others. Effect sizes ranged between 50% and 80% of a standard deviation –
13
We found similar results regardless of whether a stated-preference (i.e. importance for meeting new
people and making new friends) or a revealed-preference item (i.e. frequency of meeting in local clubs) was used
to measure the importance of social context to participants.
14
Results are available upon request.
15
Data on these additional outcomes had only been collected at a later stage (starting from t=1).
comparable to our main outcomes.16 Impacts of mindfulness on wellbeing and the importance
of social relationships and pro-social behaviour for wellbeing are well-documented in the lit-
erature (see Bohlmeijer et al. (2010), Godfrin and van Heeringen (2010), or Gu et al. (2015)
for mindfulness; Powthavee (2008) for social relationships; or Borgonovi (2008), Meier and
Stutzer (2008), or Dolan et al. (2021) for pro-social behaviour, for example).
To shed light on whether some participants benefited more than others, we conducted
a heterogeneity analysis, running separate regressions for participants in different terciles of
the respective self-reported outcome distribution, pre-treatment.17 Figure 1 in the Supplemen-
tary Materials shows our findings: only in case of PHQ-9 scores did differences between
terciles turn out to be significant. Impacts on participants in the first tercile of PHQ-9 scores
(who were more depressed) were almost seven times larger than for those in the bottom tercile
(who were less); the difference was significant at the 5% level. Besides that, we did not find
much evidence for heterogeneous effects.
Impacts on Cortisol and Cytokines. We next look at biomarkers – cortisol as a stress
response hormone and a range of cytokines as immune response proteins associated with
mental ill health and depressive symptoms. Figure 4 shows coefficient estimates.
Figure 4
Impacts on Biomarkers: Cortisol and Cytokines
16
See Supplementary Materials Tables 3a and 3b for these findings.
17
The choice of terciles was motivated by sample size.
Notes: Outcomes have been standardised prior to running regressions (i.e. transformed to z-scores with mean of
zero and standard deviation of one, using the control group mean and standard deviation). See Supplementary
Materials Table 2b for the corresponding regression table with controls. Robust standard errors were clustered at
the participant level. N between 236 and 275 depending on biomarker due to removal of outliers. Confidence
bands are 95%.
We did not find that the course had significant impacts on biomarkers at conventional levels.
However, we found that cytokines consistently moved into the hypothesised direction: pro-
inflammatory cytokines IL-1 and IL-6, which correlate positively with depressive symptoms,
decreased, whereas anti-inflammatory cytokine IL-10, interferon IFN- , and chemokine IL-8
(which correlate negatively) increased. Compared to self-reported outcomes, biomarkers were
noisier and impacts smaller in size.18
18
As with our self-reported outcomes, we ran separate regressions for participants in different terciles
of the respective biomarker distribution, pre-treatment. Figure 2 in the Supplementary Materials plots coefficient
estimates: we found again little systematic evidence that the course had significant impacts by tercile at conven-
tional levels.
Robustness. To the extent that out-of-sample selection was not random and correlated
with outcomes (for example, unhappier people may have been more likely to drop out of the
study), or differed by group, it would have biased our identified effects. We looked at this at-
trition by regressing the number of periods on each outcome alongside course-set-specific
fixed effects, using robust standard errors clustered at the individual level. We found little evi-
dence that outcomes were significant predictors of the number of periods participants re-
mained in the programme, neither on average nor by group.19 We take this as evidence that
out-of-sample selection was rather random. Note that only about 5% of participants dropped
out between t=0 and t=1, and a slightly larger proportion (22%) between t=1 and t=2. Finally,
compliance was high: on average, participants attended seven out of eight sessions.
Replication
Since its launch in 2015, 431 courses have been completed worldwide, totalling 5,621 partici-
pants. From the beginning, the charity running the courses – Action for Happiness – has been
collecting data on course outcomes at the participant level. Participants are sent a link to the
survey at t=0 after registering online for the course. Completing the online survey is manda-
tory for course participation. After the course has finished, they are again sent a link to the
survey at t=1, whereby completion is incentivised by a voucher for a free, one-year subscrip-
tion to a mindfulness app.
In particular, by means of online surveys, data on course participants' life satisfaction,
mental wellbeing, compassion, and social trust have been collected. Mental wellbeing is
measured using the Short Warwick-Edinburgh Mental Well-being Scale, which asks respond-
ents to report the frequency of several experiences related to their mental wellbeing during the
19
past two weeks. The item is bound between seven and 35, whereby higher scores indicate
higher mental wellbeing.
Although a before-after comparison of these measures does not yield causal effects of
course participation on course outcomes, we can still use these online surveys, which are
high-powered and widely spread across geographical regions and over time, to check the ex-
ternal validity of our main findings, which were based on six courses in London between
2016 and 2017. Figure 5 shows the results of this before-after comparison of course outcomes
collected via online surveys, restricted to respondents for whom we had both data at t=0 and
t=1, amounting to about 5,600 individuals (about 2,300 observations before and 2,300 after)
for comparison.
Figure 5
Impacts on Self-Reported Outcomes in Online Surveys: Life Satisfaction, Mental Wellbeing,
Compassion, and Social Trust

Notes: Data at t=0 and t=1 from online surveys on the universe of courses during the period 2015 to 2019.
Scores are in natural units. Life satisfaction and social trust were measured on scales from zero to ten; mental
wellbeing by means of the Short Warwick-Edinburgh Mental Well-being Scale, which runs from seven to 35;
and compassion by means of the Santa Clara Brief Compassion Scale, which runs from five to 35. Confidence
intervals are 95%.
Similar to the findings in our trial, the before-after comparison showed strong, positive associ-
ations between course completion and life satisfaction, mental wellbeing, compassion, and so-
cial trust.
Associations were, however, larger: for life satisfaction, for example, we found a mean
difference of about 1.4 points on a zero-to-ten scale (pre-mean of 6.1, post-mean of 7.5).
Larger associations could be driven by three factors: first, our before-after comparison did not
account for general trends in wellbeing. Second, larger associations could, in part, be driven
by attrition in online surveys: whereas attrition was low in our trial (only about 5% of partici-
pants dropped out between t=0 and t=1), attrition in online surveys was much higher, at about
36%. Finally, larger impacts could be explained by the timing of surveys at t=1: the link to the
survey is sent out shortly after the course has finished, whereas in our trial data at t=1 had
been collected before the last session started. It is therefore possible that participants' euphoria
of having finished the course was mixed up with actual course outcomes in online surveys.
Discussion
Using a randomised controlled trial, we found that the "Exploring What Matters" course had
strong, positive causal effects on participants' self-reported subjective wellbeing and mental
health. It also induced a shift in participants' attitudes towards more pro-sociality. These im-
pacts seemed to be sustained at t=2 two months post-treatment. An analysis of the mecha-
nisms of wellbeing change suggested that effects on participants may have come about
through changes in knowledge of wellbeing and behaviour in areas that have been shown to
be important for wellbeing and in which there is little hedonic adaptation, including mindful-
ness, social relationships, and pro-social behaviour. Biomarkers collected through saliva sam-
ples, including cortisol and a range of cytokines involved in inflammatory response, moved
consistently into the hypothesised direction yet failed to reach statistical significance at con-
ventional levels.
One explanation for why we did not find significant effects on biomarkers may be
power issues combined with relatively noisy measures. Another, related explanation may be
the composition of our sample: high levels of pro-inflammatory cytokines have been found
for major depression. Respondents in our sample, however, reported only mild depressive
symptomatology on average, pre-treatment. In fact, we found that only eight out of 133 re-
spondents for whom we had data at t=1 (about 6%) reported strong symptomatology, as indi-
cated by PHQ-9 scores of fifteen or higher. Moreover, even amongst these, only about a third
showed associated elevated inflammation (Wium-Andersen and Nielsen, 2013). For cortisol,
individual differences and timing of measurement matter: it has been found to be a rather
short-term measure for stress (Miller et al., 2007). Another, complementary explanation is
that the course improves participants' positivity towards life more generally, which is initially
captured by self-reported outcomes and may manifest itself in impacts on biomarkers only in
the long-run. Indeed, there is some indication in the literature that tangible health outcomes of
wellbeing interventions are attainable only in the longer term, especially if participants are
motivated to sustain the behaviour promoted during the intervention afterwards, possibly over
a period of months (see Steptoe (2019) for a review). While effects on biomarkers turned out
insignificant, the fact that they consistently moved into the hypothesised direction still sug-
gests a promising avenue for future exploration amongst individuals specifically with higher
levels of depressive symptoms at t=0, and in particular, for long-run follow-up measurement.
Compared to the literature, impacts on self-reported outcomes were large: the course
increased participants' life satisfaction on a zero-to-ten scale by about one point, which is
more than being partnered as opposed to being single (+0.6) (Clark et al., 2018). Impacts were
stronger than those found in trials funded by the UK Big Lottery Fund, which financed a wide
range of wellbeing programmes (fourteen portfolios, each consisting of three to 34 actual tri-
als) from 2008 to 2015 at a volume of £200 ($251) million. Trials typically included commu-
nity-based activities such as community gardens or sports events. As a conservative estimate,
they increased life satisfaction on a zero-to-ten scale by, on average, 0.5 points for six months
post-treatment (New Economics Foundation-Centre for Local Economic Strategies, 2013).
Different from the "Exploring What Matters" course, however, these trials targeted specific
groups with mental health needs.
Finally, impacts were highly comparable to those of ENHANCE (for life satisfaction,
about one point in "Exploring What Matters" versus 1.1 points in ENHANCE), a 12-week
wellbeing course focusing primarily on positive habits, skills, and attitudes, which is the most
comparable intervention and which can be delivered both offline and online, with little
reported differences between both delivery modes (Heintzelman et al., 2019).20 The authors
were able to provide evidence of positive impacts over a period of up to six months post-treat-
ment. Note that this six-month post-treatment period includes a three-month sub-period in
which participants who had finished the course were repeatedly followed up: in the offline
version, this included an alternating series of biweekly phone calls (of ten to fifteen minutes
duration each) and in-person group sessions (of two hours duration each) during these three
months; participants in the online version received six bi-weekly e-mails during this period.
"Exploring What Matters" differs from ENHANCE in several, fundamental aspects.
We limit our comparisons to the offline version of ENHANCE because there exists, to date,
no online version of "Exploring What Matters".21 Different from ENHANCE, "Exploring
What Matters" is led exclusively by non-expert volunteers (essentially laypeople), whereas
ENHANCE relies on graduate-level trained clinicians. This is interesting, because it shows
that laypeople without any specific academic background can be effectively utilised to sys-
tematically improve the wellbeing and pro-sociality of others. In fact, the manualisation of the
"Exploring What Matters" course and its reliance on volunteer laypeople as course leaders
make it highly cost-effective for face-to-face settings: costing only £90 ($113) per WELLBY
(a one-point increase in life satisfaction on a zero-to-ten scale for one individual for one year),
it is well above the advocated wellbeing cost-effectiveness threshold of about £2,500 ($3,139)
derived from marginal National Health Service (NHS) spending in the UK (Clark et al.,
2018), and well below the individual willingness to pay for one WELLBY of about £9,000
20
The impact of this course has been studied using a waitlist randomisation design, as in our paper, and
the authors found an impact of about 0.5 between baseline and posttest on life satisfaction measured on a one-to-
five multi-item summed scale (the Satisfaction With Life Scale) (Heintzelman et al., 2019, Table 3). With the
caveat that both measures of life satisfaction are not perfectly comparable, rescaling this item to a zero-to-ten
scale yields an impact of about 0.5*(11/5)=1.1.
21
In light of Covid-19, Action for Happiness, the charity running the "Exploring What Matters" course,
has developed a new version of the course optimised for online delivery, due to be launched in 2021. During the
pandemic, over 100 local groups have conducted the course online using Zoom and over 5,000 participants have
been involved.
($11,300) derived from marginal health improvements (Huang et al., 2018).22, 23 Another fun-
damental difference between the two courses is the period after the course has ended. Differ-
ent from ENHANCE, "Exploring What Matters" includes no three-month, labour-intensive
maintenance period (i.e. biweekly alternating phone calls and group sessions), which should
be seen as part of the intervention package and which has implications for cost-effectiveness.
Such a period may not be necessary, considering the similarity in outcomes between the two
courses.
Regardless of these differences, ENHANCE and the "Exploring What Matters" course
show remarkable similarities in terms of impacts and demonstrate that the wellbeing of
healthy adults in the general population can be effectively improved by means of intervention.
An important, policy-relevant question is how average people can be motivated to take up
wellbeing interventions (for example, by targeting their expectancies or subjective valuations
of interventions), especially if they may not believe ex-ante in their effectiveness and inter-
ventions may therefore represent credence goods, i.e. goods of which the value only becomes
apparent upon consumption.
Our study has several shortcomings. The most important one is that significant effects
on self-reported outcomes were not mirrored by biomarkers. Impacts at t=1 may thus have re-
flected participant's euphoria of having finished the course, placebo effects, or social desira-
bility if course participants tried to please course leaders. Although none of these can be ruled
out for sure, we argue that it is unlikely that our impacts were primarily driven by these
22
The wellbeing cost-effectiveness threshold of about £2,500 derived from marginal NHS spending can
be calculated as follows: the NHS approves treatment if the QALY per cost ratio is 1/£25,000. Since QALYs are
measured on a scale from zero-to-one and life satisfaction is measured on a scale from zero-to-ten, the translated
advocated wellbeing cost-effectiveness threshold becomes (1/£25,000)*10. See Layard (2016), Clark et al.
(2018), and Frijters et al. (2020) for the concept of WELLBY and Frijters and Krekel (2021) for a discussion of
wellbeing cost-effectiveness analysis.
23
We made the assumption that sustained impacts are sustained for at least one year. If we assume that
they are sustained for two months only, for which we have suggestive evidence, the course would cost £540
($678) per WELLBY.
artefacts. First, recall that the atmosphere during data collection (including t=0, t=1, and t=2)
was kept strictly neutral according to protocol, and that participants could meet and chat to
each other only after data collection had finished. Second, there was evidence for sustained
impacts: it is unlikely that placebo effects were sustained two months post-treatment. Moreo-
ver, impacts at t=2 were similar (if not stronger) than at t=1: it is unlikely that, two months af-
ter having completed the last survey, participants perfectly recalled their previous responses.
Likewise, the fact that different types of self-reported outcomes, particularly, PHQ-9 and
GAD-7, point into the same direction makes us more confident in that our identified treatment
effects are not driven exclusively by demand effects. Arguably, PHQ-9 and GAD-7 should be
less susceptible to such effects, because they (a) are multi-item summary scales (and hence
relatively less prone to them), (b) ask about actual experiences during the past two weeks (for
example, trouble falling or staying asleep, or sleeping too much), and (c) are not framed
around the notion of "happiness" (which the course is advertised to promote). Finally, data
collection was strictly anonymous, and there was little incentive for participants to answer in a
strategic or socially desirable way. Likewise, anonymous online surveys from the universe of
courses conducted showed similar impacts. They also pointed against observer effects: for
participants who completed online surveys, no field experiment was salient.
Despite these protocols, two other types of placebo effects are thinkable: first, partici-
pants self-selected into the intervention (i.e. knowing that it aims at increasing their happi-
ness) and were likely to be actively looking to improve their lives. The question then arises
whether our identified treatment effects are due to placebo effects (i.e. motivated cognition) to
which self-selected participants may be especially susceptible. Alternatively, one might argue
that self-selected participants may be especially motivated to "work hard" in order to improve
their lives, i.e. pure motivational effects. Unfortunately, our study design does not allow us to
disentangle these effects, but the literature provides evidence on their relative importance.
Lyubomirsky et al. (2011) show that self-selection strengthens treatment effects, but only
when interacted with a wellbeing-enhancing treatment (as opposed to a neutral control).
Moreover, the authors show that self-selected participants put more effort into treatment com-
pared to non-self-selected participants. Hence, self-selection seems to matter, but not so much
because of motivated cognition. Rather, it seems that self-selected participants bring with
them more positive behavioural attitudes towards treatment.
A second placebo effect may arise from the upfront donation (£90) that participants
may make in order to cover costs: one could argue that, because participants paid upfront,
they may report a higher wellbeing ex-post due to cognitive dissonance. Although we cannot
fully exclude this possibility, the combination of (i) the rather small amount (i.e. between £90
/ (8*2) = £5.6 to £90 / (10*2) = £4.5 per course hour for a course duration of between 16 and
20 hours); (ii) the relatively long duration between payment and outcome measurement of
more than two months; and (iii) the fact that course participants were not primarily from the
lower end of the income distribution reduces the likelihood of significant placebo effects from
the possibility of making an upfront donation.
Another shortcoming was the waitlist randomisation design: the choice of this design
was motivated by the fact that – in our non-clinical, general adult population, and local com-
munity intervention context – there exists no natural, credible control group that could lend
itself as a counterfactual business-as-usual. At the same time, alternate double-blind impact
study designs with placebo control groups are difficult to implement in the context of course-
based social-psychological interventions (Herbert and Gaudiano, 2005). On the one hand, a
placebo (for example, having meetings at the same time as the treatment group but in an un-
structured format without delivering course contents) could have helped to better isolate and
identify the active ingredients of the intervention (for example, specific course contents versus
socialising or disrupting the daily routine), beyond the self-reported changes in information
and behaviour that we document. On the other hand, a placebo that eliminates (ideally) one
specific channel is difficult to find and implement, especially in case of in-person courses
involving several sessions over a long period of time. Ideally, one would want to work with
multiple control groups and placebos, which can easily become quite complex. At a concep-
tual level, this raises the question of whether or not elements like socialising or disrupting the
daily routine are themselves active parts of the intervention package.
Our waitlist randomisation, therefore, balanced these challenges while adhering as
closely as possible to evidence-based practice in social science. Nevertheless, it has draw-
backs. The most important one is that being waitlisted itself could be a treatment. Bias could
have gone both ways. We found little evidence for either: between t=0 and t=1, there were lit-
tle significant differences in outcomes and covariates for the waitlisted control group, except
for mindfulness and meditation (which the waitlisted control group seemed to practice more at
t=0). Excluding individuals for whom this behavioural change occurred between t=0 and t=1
left our findings unchanged.24
Future research may build on and extend the evidence established in this trial, for ex-
ample, by looking at long-term impacts that go beyond two months post-treatment. Moreover,
it may be interesting to look at behavioural spillovers from one life domain to another or well-
being spillovers between individuals. We found participants who were initially in more men-
tal distress to benefit more from the course. A larger sample size could help stratifying results
by demographics and other participant characteristics, providing useful insights into targeting
particular groups of people more effectively. It may also help resolve power issues with bi-
omarkers. Finally, motivated by the growing literature on mentoring and advice-giving in so-
cial psychology rooted in self-perception theory and advocacy, studying the causal effect of
the course on the wellbeing of facilitators (i.e. the volunteers who lead the course) would be a
promising avenue for future research.
24
Conclusion
Our study shows that wellbeing is not fixed but can be changed by means of intervention,
cost-effectively and at scale, and that self-reported impacts are sustained over time. In particu-
lar, exposing people to the scientific evidence base on what has been found to cause wellbeing
(even when presented by non-expert laypeople), jointly discussing this evidence, and commit-
ting to make behavioural changes to daily routines can have lasting impacts on wellbeing.
This speaks against a set point of wellbeing around which individuals fluctuate and return to
by adapting to changes in life circumstances (Brickman and Campbell, 1971). Rather, the evi-
dence presented here speaks for an expectancy-value approach to behaviour change (Battle,
1965), in which individuals – once they update their beliefs about what matters to their well-
being, change their behaviour initially, and experience an initial increase in wellbeing – may
change their behaviour more permanently, with then sustained impacts on wellbeing. To the
extent that people do not anticipate or believe in such interventions, these may constitute cre-
dence goods and there may be a role to play for policy to accredit their effectiveness and dis-
seminate that information.
This has important implications for economics: apart from wellbeing being a signifi-
cant predictor of economic behaviour and individual-level outcomes such as productivity and
income (De Neve and Oswald, 2012; Oswald et al., 2015; Béllet et al., 2020), health (Graham
and Pinto, 2019), voting (Liberini et al., 2017), or organisation-level productivity and profita-
bility (Krekel et al., 2019), there are important implications for measuring societal progress
more generally. If wellbeing is not fixed and adaptation is not inevitable (e.g. we know that
there is no full adaptation to unemployment, cf. Clark et al., 2008), wellbeing can be used as a
meaningful indicator to measure societal progress, and help direct policy attention towards ar-
eas in which there may be little adaptation (such as lack of social relationships, unemploy-
ment, lack of community cohesion and trust, or mental health), and by the same token, to-
wards more wellbeing-improving activities.

The Easterlin Paradox (Easterlin, 1974, 2019) shows that, despite substantial increases
in GDP per capita, wellbeing has been largely stagnant in many developed countries over the
past decades, or even declined for some population groups (Stevenson and Wolfers, 2009).
The finding that wellbeing can improve when redirected towards certain behaviours, com-
bined with the growing evidence base on its causes and consequences, underlines its useful-
ness as an indicator for measuring how we are doing as a society, which is a core activity of
the economics profession.

Acknowledgements
We are heavily indebted to Fulvio D'Acquisto and Martin Gross at the William Harvey Re-
search Institute, Queen Mary University of London, for helping us with the logistics of tem-
porarily storing our biomarker samples. We are thankful to Ed Diener, Carol Graham, Paul
Frijters, Claryn Kung, and Ashley Whillans, as well as seminar participants at the London
School of Economics and various other places, for helpful comments and suggestions. Lucía
Macchia and Ekaterina Oparina provided excellent research assistance. A special thanks goes
to the volunteer course leaders at Action for Happiness, course participants, and to the Action
for Happiness staff, in particular Keith Cowley, Alex Nunn, and Mark Williamson. We thank
the editor, associate editor, and two anonymous referees for their very helpful comments and
suggestions.
Funding
Funding from the John Templeton Foundation and the What Works Centre for Wellbeing's
evidence programme is gratefully acknowledged.

References
Aknin, L.B., Barrington-Leigh, C.P., Dunn, E.W., Helliwell, J.F., Burns, J., Biswas-Diener,
R., Kemeza, I., Nyende, P., Ashton-James, C.E., and Norton, M.I. (2013). Pro-Social
Spending and Well-Being: Cross-Cultural Evidence for a Psychological Universal.
Journal of Personality and Social Psychology, 104(4), 635-652.
Battle, E. S. (1965). Motivational Determinants of Academic Task Persistence. Journal of
Personality and Social Psychology, 2(2), 209-2018.
Bellet, C. S., De Neve, J.-E., and Ward, G. (2020). Does Employee Happiness Have an Im-
pact on Productivity? CEP Discussion Paper, 1655.
Bohlmeijer, E., Prenger, R., Taal, E., and Cuijpers, P. (2010). The effects of mindfulness-
based stress reduction therapy on mental health of adults with a chronic medical dis-
ease: A meta-analysis. Journal of Psychosomatic Research, 68(6), 539-44.
Bolier, L., Haverman, M., Westerhof, G. J., Riper, H., Smit, F., and Bohlmeijer, E. (2013).
Positive psychology interventions: A meta-analysis of randomized controlled studies.
BMC Public Health, 13, 119.
Borgonovi, F. (2008). Doing well by doing good. The relationship between formal volunteer-
ing and self-reported health and happiness. Social Science & Medicine, 66, 2321-2334.
Brown, K. W., and Ryan, R. M. (2003). The benefits of being present: Mindfulness and its
role in psychological well-being. Journal of Personality and Social Psychology, 84,
822-848.
Brickman, P., and Campbell, D. (1971). Hedonic Relativism and Planning the Good Society.
In Apley, M. H. (ed). Adaptation-Level Theory: A Symposium. New York: Academic
Press.
Carter, T. J., and Gilovich, T. (2010). The Relative Relativity of Material and Experiential
Purchases. Journal of Personality and Social Psychology, 98(1), 146-159.

Chirkov, V. I., Ryan, R. M., Kim, Y., and Kaplan, U. (2003). Differentiating autonomy from
individualism and independence: A self-determination theory perspective on internali-
zation of cultural orientations and well-being. Journal of Personality and Social Psy-
chology, 84, 97-110.
Church, A. T., Katigbak, M. S., Locke, K. D., Zhang, H., Shen, J., Jesús de Vargas-Flores, J.,
Tanaka-Matsumi, J., et al. (2013). Need Satisfaction and Well-Being: Testing Self-De-
termination Theory in Eight Cultures. Journal of Cross-Cultural Psychology, 44(4),
507-534.
Clark, A. E., Diener, E., Georgellis, Y., and Lucas, R. E. (2008). Lags And Leads in Life Sat-
isfaction: a Test of the Baseline Hypothesis. Economic Journal, 118(529), F222-F243.
Clark, A. E., Flèche, S., Layard, R., Powdthavee, N., and Ward, G. (2018). The Origins of
Happiness: The Science of Well-Being Over the Life Course. Princeton, NJ: Princeton
University Press.
Clark, D. M., Layard, R., Smithies, R., Richards, D. A., Suckling, R., and Wright, B. (2009).
Improving access to psychological therapy: Initial evaluation of two UK demonstra-
tion sites. Behaviour Research and Therapy, 47(11), 910-920.
Danner, D. D., Snowdon, D. A., and Friesen, W. V. (2001). Positive Emotions in Early Life
and Longevity: Findings from the Nun Study. Journal of Personality and Social Psy-
chology, 80(5), 804-813.
Deci, E. L., and Ryan, R. M. (1985). Intrinsic motivation and self-determination in human be-
haviour. New York: Plenum.
De Neve, J.-E., and Oswald, A. J. (2012). Estimating the influence of life satisfaction and pos-
itive affect on later income using sibling fixed effects. Proceedings of the National
Academy of Sciences, 109(49), 19953-19958.

Di Tella, R., Haisken-DeNew, J., and MacCulloch, R. (2010). Happiness Adaptation to In-
come and to Status in an Individual Panel. Journal of Economic Behaviour & Organi-
zation, 76, 834-852.
Diener, E., Suh, E. M., Lucas, R. E., and Smith, H. L. (1999). Subjective Well-Being: Three
Decades of Progress. Psychological Bulletin, 125(2), 276-302.
Dolan, P., Krekel, C., Shreedhar, G., Lee, H., Marshall, C., and Smith, A. (2021). Volunteer-
ing Improves Wellbeing: Evidence from a Nationwide Micro-Volunteering Pro-
gramme. CEP Discussion Paper.
Dowlati, Y., Herrmann, N., Swardfager, W., Liu, H., Sham, L., Reim, E. K., and Lanctôt, K.
L. (2010). A Meta-Analysis of Cytokines in Major Depression. Biological Psychiatry,
67, 446-457.
Drouvelis, M., and Grosskopf, B. (2016). The effects of induced emotions on pro-social be-
haviour. Journal of Public Economics, 134, 1-8.
Dunn, E.W., Aknin, L.B., and Norton, M.I. (2008). Spending Money on Others Promotes
Happiness. Science, 319, 1687-1688.
Easterlin, R. (1974). Does Economic Growth Improve the Human Lot? Some Empirical Evi-
dence. In: David, R., and Reder, R. (eds). Nations and Households in Economic
Growth: Essays in Honor of Moses Abramovitz. New York: Academic Press.
Easterlin, R. A., Angelescu McVey, L., Switek, M., Sawangfa, O., and Smith Zweig, J.
(2010). The happiness-income paradox revisited. Proceedings of the National Acad-
emy of Sciences, 107(52), 22463-22468.
Fancourt, D., Perkins, R., Ascenso, S., Atkins, L., Kilfeather, S., Carvalho, L., Steptoe, A.,
and Williamon, A. (2016). Group Drumming Modulates Cytokine Response in Mental
Health Services Users: A Preliminary Study. Psychotherapy and Psychosomatics,
85(1), 53-55.
Frederick, S., and Loewenstein, G. (1999). Hedonic Adaptation. In Kahneman, D., Diener, E.,
and Schwartz, N. (eds). Well-Being: The Foundations of Hedonic Psychology. New
York: Russell Sage.
Frijters, P., and Krekel, C. (2021). A Handbook for Wellbeing Policy-Making. Oxford: Oxford
University Press.
Frijters, P., Clark, A. E., Krekel, C., and Layard, R. (2020). A happy choice: wellbeing as the
goal of government. Behavioural Public Policy, 4(2), 126-165.
Graham, C., and Pinto, S. (2019). Unequal hopes and lives in the USA: Optimism, race, place,
and premature mortality. Journal of Population Economics, 32(2), 665-733.
Godfrin, K. A., and van Heeringen, C. (2010). The effects of mindfulness-based cognitive
therapy on recurrence of depressive episodes, mental health and quality of life: A ran-
domized controlled study. Behaviour Research and Therapy, 48(8), 738-746.
Gu, J., Strauss, C., Bond, R., and Cavanagh, K. (2015). How do mindfulness-based cognitive
therapy and mindfulness-based stress reduction improve mental health and wellbeing?
A systematic review and meta-analysis of mediation studies. Clinical Psychology Re-
view, 37(1), 1-12.
Hausman, L. R. M., Parks, A., Youk, A. O., and Kwoh, C. K. (2014). Reduction of bodily
pain in response to an online positive activities intervention. Journal of Pain, 15, 560-
567.
Heintzelman, S. J., Kushlev, K., Lutes, L. D., Wirtz, D., Kanippayoor, J. M., Leitner, D., Oi-
shi, S., and Diener, E. (2019). ENHANCE: Evidence for the efficacy of a comprehen-
sive intervention program to promote subjective well-being. Journal of Experimental
Psychology: Applied, forthcoming.
Herbert, J. D., and Gaudiano, B. A. (2005). Moving from empirically supported treatment lists
to practice guidelines in psychotherapy: The role of the placebo concept. Journal of
Clinical Psychology, 61(7), 893-908.

Huang, L., Frijters, P., Dalziel, K., and Clarke, P. (2018). Life satisfaction, QALYs, and the
monetary value of health. Social Science & Medicine, 211, 131-136.
Kahneman, D. (2000). Experienced Utility and Objective Happiness: A Moment-Based Ap-
proach. In Kahneman, D., and Tversky, A. (eds). Choice, Values, and Frames. New
York: Cambridge University Press.
Krekel, C., Swanke, A., De Neve, J.-E., and Fancourt, D. (2020). Are Happier People More
Compliant? Global Evidence From Three Large-Scale Surveys During Covid-19
Lockdowns. IZA Discussion Paper, 13690.
Krekel, C., Ward, G., and De Neve, J.-E. (2019). Employee Wellbeing, Productivity and Firm
Performance. CEP Discussion Paper, 1605.
Kuhn, P., Kooreman, P., Soetevent, A., and Kapteyn, A. (2011). The Effects of Lottery Prizes
on Winners and Their Neighbors: Evidence from the Dutch Postcode Lottery. Ameri-
can Economic Review, 101(5), 2226-2247.
La Guardia, J. G., Ryan, R. M., Couchman, C. E., and Deci, E. L. (2000). Within-person vari-
ation in security of attachment: A self-determination theory perspective on attachment,
need fulfillment, and well-being. Journal of Personality and Social Psychology, 79(3),
367-384.
Layard, R. (2016). Wellbeing measurement and cost-effectiveness analysis. CEP Working Pa-
per.
Layard, R., Clark, A. E., Cornaglia, F., Powdthavee, N., and Vernoit, J. (2014). What Predicts
a Successful Life? A Life‐course Model of Well‐being. Economic Journal, 124(580),
F720-F738.
Liberini, F., Redoano, M., and Proto, E. (2017). Happy voters. Journal of Public Economics,
146, 41-57.
Lucas, R. (2005). Time Does Not Heal All Wounds: A Longitudinal Study of Reaction and
Adaptation to Divorce. Psychological Science, 16(12), 945-950.

Lucas, R., and Clark, A. E. (2006). Do People Really Adapt to Marriage? Journal of Happi-
ness Studies, 7(4), 405-426.
Lyubomirsky, S., Dickerhoof, R., Boehm, J. K., and Sheldon, K. M. (2011). Becoming Hap-
pier Takes Both a Will and a Proper Way: An Experimental Longitudinal Intervention
To Boost Well-Being. Emotion, 11(2), 391-402.
Meier, S., and Stutzer, A. (2008). Is Volunteering Rewarding in Itself? Economica, 75(297),
39-59.
Menzel, P., Dolan, P. H., Richardson, J., and Olsen, J. (2002). The role of adaptation to disa-
bility and disease in health state valuation: a preliminary normative analysis. Social
Science & Medicine, 55(12), 2149-2158.
Milyavskaya, M., and Koester, R. (2011). Psychological needs, motivation, and well-being: A
test of self-determination theory across multiple domains. Personality and Individual
Differences, 50(3), 387-391.
Miller, A. H., and Raison, C. L. (2016). The role of inflammation in depression: from evolu-
tionary imperative to modern treatment target. Nature Reviews Immunology, 16(1), 22.
Miller, G. E., Chen, E., and Zhou, E. S. (2007). If It Goes Up, Must It Come Down? Chronic
Stress and the Hypothalamic-Pituitary-Adrenocortical Axis in Humans. Psychological
Bulletin, 133(1), 25-45.
New Economics Foundation – Centre for Local Economic Strategies (2013). Big Lottery Fund
National Well-Being Evaluation. Draft Report.
NHS Long Term Plan (2019). The NHS Long Term Plan 2019. Online: https://www.long-
termplan.nhs.uk/wp-content/uploads/2019/08/nhs-long-term-plan-version-1.2.pdf, last
accessed 20/04/2021.
Oswald, A. J., and Gardner, J. (2006). Do divorcing couples become happier by breaking up?
Journal of the Royal Statistical Society Series A: Statistics in Society, 169(2), 319-336.
Oswald, A. J., and Powdthavee, N. (2008). Does Happiness Adapt? A Longitudinal Study of
Disability with Implications for Economists and Judges. Journal of Public Economics,
92(5-6), 1061-1077.
Oswald, A. J., Proto, E., and Sgroi, D. (2015). Happiness and Productivity. Journal of Labor
Economics, 33(4), 789-822.
Powdthavee, N. (2008). Putting a price tag on friends, relatives, and neighbours: Using sur-
veys of life satisfaction to value social relationships. Journal of Socio-Economics,
37(4), 1459-1480.
Querstret, D., Morison, L., Dickinson, S., Cropley, M., and John, M. (2020). Mindfulness-
based stress reduction and mindfulness-based cognitive therapy for psychological
health and well-being in nonclinical samples: A systematic review and meta-analysis.
International Journal of Stress Management, 27(4), 394-411.
Romano, J. P., and Wolf, M. (2005a). Stepwise Multiple Testing as Formalized Data Snoop-
ing. Econometrica, 73(4), 1237-1282.
Romano, J. P., and Wolf, M. (2005b). Exact and Approximate Stepdown Methods for Multi-
ple Hypothesis Testing. Journal of the American Statistical Association, 469(100), 94-
108.
Romano, J. P., and Wolf, M. (2016). Efficient computation of adjusted p-values for
resampling-based stepdown multiple testing. Statistics & Probability Letters, 113, 38-
40.
Ryan, R. M., and Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic
motivation, social development, and well-being. American Psychologist, 55(1), 68-78.
Sedlmeier, P., Eberth, J., Schwarz, M., Zimmermann, D., Haarig, F., Jaeger, S., and Kunze, S.
(2012). The psychological effects of meditation: A meta-analysis. Psychological Bul-
letin, 138(6), 1139-1171.

Sin, N. L., and Lyubomirsky, S. (2009). Enhancing well-being and alleviating depressive
symptoms with positive psychology interventions: A practice-friendly meta-analysis.
Journal of Clinical Psychology, 65(5), 467-487.
Steptoe, A. (2019). Happiness and Health. Annual Review of Public Health, 40, 339-59.
Steptoe, A., and Wardle, J. (2011). Positive affect measured using ecological momentary as-
sessment and survival in older men and women. Proceedings of the National Academy
of Sciences, 108(45), 18244-18248.
Stevenson, B., and Wolfers, J. (2009). The Paradox of Declining Female Happiness. Ameri-
can Economic Journal: Economic Policy, 1(2), 190-225.
Stutzer, A., and Frey, B. (2006). Does marriage make people happy, or do happy people get
married? Journal of Socio-Economics, 35(2), 326-347.
Taylor, C. T., Lyubomirsky, S., and Stein, M. B. (2017). Upregulating the positive affect sys-
tem in anxiety and depression: Outcomes of a positive activity intervention. Depres-
sion and Anxiety, 34, 267-280.
Tomitaka, S., Kawasaki, Y., Ide, K., Akutagawa, M., Ono, Y., and Furukawa, T. A., (2018).
Stability of the Distribution of Patient Health Questionnaire-9 Scores Against Age in
the General Population: Data From the National Health and Nutrition Examination
Survey. Frontiers in Psychology, 9, 390.
Wiles, N., Thomas, L., Turner, N., Garfield, K., Kounali, D., Campbell, J., Kessler, D., et al.
(2016). Long-term effectiveness and cost-effectiveness of cognitive behavioural ther-
apy as an adjunct to pharmacotherapy for treatment-resistant depression in primary
care: follow-up of the CoBalT randomised controlled trial. Lancet: Psychiatry, 3, 137-
144.
Wium-Andersen, M. K., and Nielsen, S. F. (2013). Elevated C-Reactive Protein Levels, Psy-
chological Distress, and Depression in 73,131 Individuals. JAMA Psychiatry, 70(2),
176-184.
Zhou, X., Fragala, M. S., McElhaney, J. E., and Kuchel, G. A. (2010). Conceptual and meth-
odological issues relevant to cytokine and inflammatory marker measurements in clin-
ical research. Current Opinion in Clinical Nutrition & Metabolic Care, 13(5), 541-
547.
Supplementary Materials
Figures
Figure 1
Impacts on Self-Reported Outcomes by Tercile: Subjective Wellbeing, Mental Health, and Pro-Sociality
Notes: Sample is split by tercile of respective outcome distribution, pre-treatment. Outcomes have been standardised prior to running regressions (i.e. transformed to z-scores
with mean of zero and standard deviation of one, using the control group mean and standard deviation). Controls include age, gender, marital status, education, employment,
income, religion, religious practice, preference for meeting new people and making friends, health (including pregnancy), health-related behaviour (including smoking and medi-
cation usage), and course-set-specific fixed effects. Robust standard errors are clustered at the participant level. N=279 (146 respondents, of which 73 are in treatment and 73 in
control). Confidence bands are 95%.
Figure 2
Impacts on Biomarkers by Tercile: Cortisol and Cytokines
Notes: Sample is split by tercile of respective outcome distribution, pre-treatment. Outcomes have been standardised prior to running regressions (i.e. transformed to z-scores
with mean of zero and standard deviation of one, using the control group mean and standard deviation). Controls include age, gender, marital status, education, employment,
income, religion, religious practice, preference for meeting new people and making friends, health (including pregnancy), health-related behaviour (including smoking and medi-
cation usage), and course-set-specific fixed effects. Robust standard errors are clustered at the participant level. N=279 (146 respondents, of which 73 are in treatment and 73 in
control). Confidence bands are 95%.
Tables
Table 1a
Comparison of Understanding Society in London with Estimation Sample
Mean
Mean Estimation Sample,
Understanding Society Difference
Pre-Treatment
(London, Same Age Span)
Subjective Wellbeing
Life Satisfaction 7.087 6.349 0.738***
Demographic Characteristics
Age: 20-24 0.101 0.055 0.046*

25-34 0.212 0.267 -0.055
35-44 0.242 0.232 0.009
45-54 0.217 0.226 -0.009
55-64 0.136 0.178 -0.042
65-74 0.092 0.041 0.051**
Gender: Male 0.547 0.171 0.376***

Female 0.453 0.829 -0.376***
Marital Status: Single 0.346 0.390 -0.044

Married 0.530 0.199 0.331***
Separated 0.023 0.034 -0.011
Divorced 0.069 0.102 -0.034
Widowed 0.028 0.007 0.021
Domestic Partner 0.003 0.247 -0.243***

(Understanding Society: Civic Partner)
Prefer not to Say - 0.021
0.142
Income: £14,999 or Less 0.182 0.137 0.045

£15,000-£29,999 0.151 0.205 -0.055*
£30,000-£44,999 0.151 0.199 -0.048
£45,000-£59,999 0.137 0.130 0.007
£60,000-£74,999 0.109 0.116 -0.007
£75,000 or More 0.270 0.157 0.112***
Prefer not to Say - 0.055
0.228
Observations 28,547 146 -

Notes: See Table Supplementary Materials Table 1b for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
Table 1b
Variable Definitions and Descriptive Statistics
Mean Standard Minimum Maximum Number Remarks

Deviation of Obser-
vations
Self-Reported Outcomes
Life Satisfaction 6.570 1.669 1 10 279 “Overall, how satisfied are you with your
life nowadays?”:
(0) “Not at all” to (10) “Completely”
Happiness 6.376 1.989 1 10 279 “Overall, how happy did you feel yester-
day?”:
Anxiousness 4.133 2.489 0 10 279 “Overall, how anxious did you feel yester-
day?”:
Worthwhileness 7.194 1.827 1 10 279 “Overall, to what extent do you feel the
things you do
in your life are worthwhile?”:
PHQ-9 (Depression) 6.358 4.523 0 21 279 9-Item Patient-Health Questionnaire,

see Kroenke et al. (2001)
GAD-7 (Anxiety) 6.057 4.640 0 20 279 7-Item Generalised-Anxiety-Disorder Ques-

tionnaire,
see Spitzer et al. (2006)
Compassion 6.762 2.398 0 11.8 279 5-Item Santa Clara Brief Compassion Scale,
see Hwang et al. (2008)
Social Trust 6.584 2.079 0 10 279 “Generally, would you say that most people
can be trusted, or that you cannot be too
careful in dealing with people?”: (0) “Can-
not be too careful” to (10) “Most can be
trusted”
Gratitude 6.222 0.890 0 7 279 “I have so much in life to be thankful for.”:
(0) “Strongly disagree” to (10) “Strongly
agree”
Information (a) 7.691 1.563 1 10 230 “I feel aware of what contributes to a happy
and meaningful life.”: (0) “Not at all” to
(10) “Completely”
Information (b) 7.374 1.738 1 10 230 “I know what really matters to me in life.”:
Information (c) 7.243 1.916 1 10 230 “I feel able to do things to improve my own
wellbeing.”: (0) “Not at all” to (10) “Com-
pletely”
Information (d) 7.274 1.602 2 10 230 “I feel able to do things to improve the well-
being of others.”: (0) “Not at all” to (10)
“Completely”
Behaviour (a) 2.057 0.897 0 3 230 “In recent weeks, how often have you done
the following? …Noticed and felt grateful
for good things”:
(0) “Not at all”, (1) “Several days”, (2)
“More than half the days”, (3) “Nearly every
day”
Behaviour (b) 1.426 1.062 0 3 230 “…Practised mindfulness/meditation”:
day”
Behaviour (c) 1.570 0.868 0 3 230 “…Treated yourself in a kind way”:
day”
Behaviour (d) 1.661 0.813 0 3 230 “…Made time for something really im-
portant to you”:
day”
Behaviour (e) 1.561 0.800 0 3 230 “…Responded well to a difficult situation”:
day”
Behaviour (f) 1.248 0.801 0 3 230 “…Learnt or tried out something new”:
day”
Behaviour (g) 1.796 0.905 0 3 230 “…Gave time to one of your closest rela-
tionships”:
day”
Behaviour (h) 1.983 0.861 0 3 230 “…Connected with other people”:
day”
Behaviour (i) 1.765 0.808 0 3 230 “…Did something kind or helpful for oth-
ers”:
day”
Behaviour (j) 1.343 0.966 0 3 230 “…Tried to increase happiness at work”:
day”
Behaviour (k) 0.896 0.845 0 3 230 “…Tried to increase happiness in the com-
munity”:
day”
Behaviour (l) 1.170 0.994 0 3 230 “…Thought about the difference you make
in the world”:
day”
Biomarkers
Cortisol 0.164 0.653 0.015 10.300 275 Cortisol in µg/dL

Interferon IFN- 7.978 26.302 0.061 205.826 243 Interferon IFN- in pg/mL
Cytokine IL-10 1.433 2.900 0.023 37.906 274 Anti-Inflammatory Cytokine IL-10 in
pg/mL
Cytokine IL-1 245.730 221.421 6.083 1,306.554 275 Pro-Inflammatory Cytokine IL-1 in pg/mL
Cytokine IL-6 9.324 18.997 0.710 206.299 275 Pro-Inflammatory Cytokine IL-6 in pg/mL
Chemokine IL-8 1,389.868 886.035 127.297 6,783.128 275 Chemokine IL-8 in pg/mL
Controls
Age: 20-24 0.050 0.219 0 1 279 -

25-34 0.265 0.442 0 1 279 -
35-44 0.233 0.423 0 1 279 -
45-54 0.237 0.426 0 1 279 -
55-64 0.176 0.381 0 1 279 -
65-74 0.039 0.195 0 1 279 -
Gender: Male 0.176 0.381 0 1 279 -
Female 0.824 0.381 0 1 279 -
Marital Status: Single 0.394 0.490 0 1 279 -
Married 0.201 0.401 0 1 279 -
Separated 0.036 0.186 0 1 279 -
Divorced 0.100 0.301 0 1 279 -
Widowed 0.007 0.085 0 1 279 -
Domestic Partner 0.247 0.432 0 1 279 -
Prefer not to Say 0.014 0.119 0 1 279 -
Educational Status: Secondary Degree 0.047 0.211 0 1 279 -
Vocational Degree 0.079 0.270 0 1 279 -
Tertiary Degree 0.384 0.487 0 1 279 -
Higher Than Tertiary Degree 0.484 0.501 0 1 279 -
Prefer not to Say 0.007 0.085 0 1 279 -
Employment Status: Working Full-Time for 0.498 0.501 0 1 279 -
Employer
Working Full-Time for Self 0.151 0.358 0 1 279 -
Working Part-Time 0.194 0.396 0 1 279 -

Working Part-Time (Underemployed) 0.004 0.060 0 1 279 -
Unemployed 0.050 0.219 0 1 279 -
Out of Labour Force 0.068 0.252 0 1 279 -
Prefer not to Say 0.036 0.186 0 1 279 -
Income: £14,999 or Less 0.143 0.351 0 1 279 -
£15,000-£29,999 0.197 0.399 0 1 279 -
£30,000-£44,999 0.201 0.401 0 1 279 -
£45,000-£59,999 0.122 0.328 0 1 279 -
£60,000-£74,999 0.122 0.328 0 1 279 -
£75,000 or More 0.161 0.368 0 1 279 -
Prefer not to Say 0.054 0.226 0 1 279 -
Religion: None 0.563 0.497 0 1 279 -
Christian 0.222 0.416 0 1 279 -
Buddhist 0.082 0.276 0 1 279 -
Hindu 0.029 0.167 0 1 279 -
Jewish 0.007 0.085 0 1 279 -
Muslim 0.007 0.085 0 1 279 -
Sikh 0.007 0.085 0 1 279 -
Other 0.029 0.167 0 1 279 -
Prefer not to Say 0.054 0.226 0 1 279 -
Religious Practice: Never 0.516 0.501 0 1 279 -
Less Than Annually 0.090 0.286 0 1 279 -
At Least Annually 0.143 0.351 0 1 279 -
At Least Monthly 0.100 0.301 0 1 279 -
At Least Weekly 0.086 0.281 0 1 279 -
Prefer not to Say 0.065 0.246 0 1 279 -
Smoking: Yes 0.082 0.276 0 1 279 -
No 0.918 0.276 0 1 279 -
Pregnant: Yes 0.014 0.119 0 1 279 -
No 0.986 0.119 0 1 279 -
Medication: Yes 0.416 0.494 0 1 279 -
No 0.584 0.494 0 1 279 -

Important to Meet New People and Make 0.799 0.401 0 1 279 “How important is it for you to be meeting
Friends: Yes new people
and making friends?”
(0) “Not at all important” to (10) “Ex-
tremely important” (=6+7+8+9+10)
No 0.201 0.401 0 1 279 (=0+1+2+3+4+5)
Table 1c
Balancing Properties Between Treatment and Control Group
Mean Mean Difference

Treatment Group, Control Group,
Pre-Intervention Pre-Intervention
Life Satisfaction 6.384 6.315 0.068

Happiness 6.151 6.288 -0.137
Anxiousness 4.233 4.438 -0.205
Worthwhileness 6.932 7.041 -0.110
PHQ-9 (Depression) 6.726 7.151 -0.425
GAD-7 (Anxiety) 6.110 6.671 -0.562
Compassion 6.523 6.792 -0.268
Social Trust 6.288 6.507 -0.219
Gratitude 6.205 6.178 0.027
Information (a) 7.479 7.469 0.010
Information (b) 7.188 7.224 -0.037
Information (c) 7.271 6.796 0.475
Information (d) 6.896 7.224 -0.329
Behaviour (a) 2.021 2.020 0.000
Behaviour (b) 1.333 1.429 -0.095
Behaviour (c) 1.417 1.449 -0.032
Behaviour (d) 1.667 1.571 0.095
Behaviour (e) 1.354 1.571 -0.217
Behaviour (f) 1.167 1.204 -0.037
Behaviour (g) 1.875 1.837 0.038
Behaviour (h) 1.854 2.061 -0.207
Behaviour (i) 1.521 1.816 -0.295*
Behaviour (j) 1.208 1.347 -0.139

Behaviour (k) 0.625 0.776 -0.151
Behaviour (l) 0.938 1.122 -0.185
Biomarkers
Cortisol 0.263 0.172 0.091

Interferon IFN- 9.613 7.510 2.103
Cytokine IL-10 1.783 1.397 0.387
Cytokine IL-1 256.544 284.884 -28.340
Cytokine IL-6 9.809 10.131 -0.322
Chemokine IL-8 1,379.952 1,603.446 -223.494
Controls
Age: 20-24 0.041 0.068 -0.027

25-34 0.288 0.247 0.041
35-44 0.233 0.233 -0.000
45-54 0.247 0.205 0.041
55-64 0.164 0.192 -0.027
65-74 0.027 0.055 -0.027
Gender: Male 0.178 0.164 0.014
Female 0.822 0.836 -0.014
Married 0.205 0.192 0.014
Separated 0.027 0.041 -0.014
Divorced 0.110 0.096 0.014
Widowed 0.000 0.014 -0.014
Domestic Partner 0.274 0.219 0.055
Prefer not to Say 0.041 0.000 0.041*
Educational Status: Secondary Degree 0.055 0.041 0.014
Vocational Degree 0.082 0.068 0.014
Tertiary Degree 0.425 0.356 0.068

Higher Than Tertiary Degree 0.425 0.534 -0.110
Prefer not to Say 0.014 0.000 0.014
Employment Status: Working Full-Time 0.534 0.466 0.068
for Employer
Working Full-Time for Self 0.137 0.164 -0.027
Working Part-Time 0.151 0.219 -0.068
Working Part-Time (Underemployed) 0.014 0.000 0.014
Unemployed 0.055 0.041 0.014
Out of Labour Force 0.055 0.096 -0.041
Prefer not to say 0.055 0.014 0.041
Income: £14,999 or Less 0.137 0.137 -0.000
£15,000-£29,999 0.219 0.192 0.027
£30,000-£44,999 0.164 0.233 -0.068
£45,000-£59,999 0.082 0.178 -0.096*
£60,000-£74,999 0.123 0.110 0.014
£75,000 or More 0.192 0.123 0.068
Religion: None 0.548 0.589 -0.041
Christian 0.192 0.247 -0.055
Buddhist 0.110 0.055 0.055
Hindu 0.041 0.014 0.027
Jewish 0.014 0.000 0.014
Muslim 0.014 0.000 0.014
Sikh 0.014 0.000 0.014
Other 0.014 0.041 -0.027
Prefer not to Say 0.055 0.055 -0.000
Religious Practice: Never 0.521 0.521 -0.000
Less Than Annually 0.068 0.110 -0.041
At Least Annually 0.164 0.123 0.041
At Least Monthly 0.082 0.110 -0.027
At Least Weekly 0.082 0.096 -0.014

Smoking: Yes 0.082 0.110 -0.027
No 0.918 0.890 0.027
Pregnant: Yes 0.000 0.027 -0.027
No 1.000 0.973 0.027
Medication: Yes 0.397 0.438 -0.041
No 0.603 0.562 0.041
Important to Meet New People and Make 0.726 0.808 -0.082
Friends: Yes
No 0.274 0.192 0.082
Observations 73 73 -
Notes: T-tests used robust standard errors clustered at individual level. See Supple-
mentary Materials Table 1b for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
Table 2a
Main Results – Self-Reported Outcomes (Regression Table for Figure 3), Including Controls
Subjective Wellbeing Mental Health Pro-Sociality

PHQ-9
Life Satis- Anxious- Worth- (Depres- GAD-7
faction Happiness ness whileness sion) (Anxiety) Compassion Social Trust Gratitude
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Treatment*Post 0.633*** 0.596*** -0.468** 0.491*** -0.497*** -0.424*** 0.383*** 0.531*** 0.286**
(0.152) (0.173) (0.190) (0.154) (0.135) (0.119) (0.145) (0.160) (0.136)
Treatment 0.0947 0.0313 0.0306 -0.0682 0.0132 -0.0622 -0.275* -0.181 0.000337
(0.168) (0.163) (0.174) (0.162) (0.168) (0.159) (0.153) (0.197) (0.172)
Post -0.0182 -0.141 0.0370 -0.0518 -0.00930 0.0680 -0.101 -0.0782 -0.0734
(0.103) (0.126) (0.134) (0.109) (0.0863) (0.0915) (0.0958) (0.102) (0.0965)
Age: 20-24
25-34 -0.437 -0.409 -0.0941 0.214 -0.464 -0.546 0.228 0.425 0.0929
(0.278) (0.330) (0.329) (0.261) (0.337) (0.374) (0.321) (0.465) (0.276)
35-44 -0.719** -0.573* 0.0531 -0.109 -0.470 -0.246 -0.263 -0.430 -0.0309
(0.318) (0.343) (0.373) (0.305) (0.337) (0.382) (0.365) (0.466) (0.328)
45-54 -0.784** -0.643* -0.289 0.131 -0.497 -0.391 -0.211 0.126 0.351
(0.355) (0.372) (0.403) (0.318) (0.348) (0.390) (0.359) (0.530) (0.324)
55-64 -0.502 -0.382 -0.209 0.00985 -0.536 -0.433 -0.271 0.292 -0.0480
(0.342) (0.374) (0.414) (0.309) (0.399) (0.393) (0.414) (0.538) (0.339)
65-74 -0.968 -0.838* -0.0171 -0.444 -0.549 -0.0981 -1.258* -0.373 -0.817
(0.594) (0.486) (0.564) (0.555) (0.511) (0.469) (0.734) (0.735) (0.513)
Gender: Male
Female 0.140 0.0599 -0.0526 0.209 -0.0558 0.0898 0.875*** 0.238 0.257
(0.209) (0.196) (0.238) (0.254) (0.220) (0.218) (0.250) (0.211) (0.188)
Marital Status: Single
Partnered 0.0655 -0.170 0.190 0.0624 0.00306 0.107 -0.103 -0.120 -0.140
(0.168) (0.172) (0.186) (0.163) (0.173) (0.184) (0.195) (0.226) (0.173)
Married 0.106 -0.282 -0.0888 -0.0236 -0.109 -0.194 0.0670 0.203 0.145
(0.214) (0.189) (0.227) (0.195) (0.204) (0.216) (0.215) (0.255) (0.202)
Separated -0.0139 -0.840** 0.0225 0.0868 0.0442 -0.0811 0.0234 -0.0862 -0.131
(0.404) (0.366) (0.381) (0.363) (0.353) (0.338) (0.405) (0.347) (0.381)
Divorced -0.446 -0.527* 0.0501 -0.756** 0.274 -0.0524 0.567* -0.502 -0.168
(0.345) (0.311) (0.301) (0.314) (0.360) (0.298) (0.294) (0.328) (0.320)
Widowed 0.550 0.109 0.646 0.685 0.463 -1.357*** 1.440* 0.639 1.445***
(0.639) (0.519) (0.582) (0.611) (0.600) (0.513) (0.759) (0.646) (0.517)
Prefer not to Say 0.832** 0.452 -0.611 -0.104 -0.460 -0.803*** -0.565 -1.120 0.721**
(0.366) (0.483) (0.511) (0.733) (0.471) (0.292) (0.506) (1.153) (0.289)
Educational Status: Secondary
Degree
Vocational Degree 0.302 0.600 -0.447 0.0765 0.268 0.240 -0.374 -0.579 -0.642
(0.547) (0.477) (0.478) (0.552) (0.506) (0.437) (0.446) (0.434) (0.642)
Tertiary Degree 0.534 0.501 -0.0735 0.246 -0.396 -0.126 -0.327 -0.595* -0.401
(0.507) (0.390) (0.362) (0.369) (0.408) (0.361) (0.252) (0.306) (0.356)
Higher Than Tertiary Degree 0.305 0.355 0.226 -0.0663 -0.00718 0.214 -0.230 -0.700** -0.513
(0.475) (0.359) (0.350) (0.354) (0.395) (0.344) (0.241) (0.304) (0.335)
Prefer not to Say 0.226 0.637 -3.393*** 0.157 -2.001*** -1.533** 0.582 -0.142 -0.396
(0.686) (0.652) (0.587) (0.651) (0.608) (0.640) (0.542) (0.726) (0.559)
Employment Status: Working
Full-Time for Employer
Working Full-Time for Self 0.0604 -0.110 0.302 0.211 0.280 0.334 0.0169 0.00683 0.0378
(0.240) (0.227) (0.250) (0.276) (0.252) (0.233) (0.306) (0.236) (0.292)
Working Part-Time 0.0458 0.183 -0.0633 -0.0667 -0.00339 -0.304 -0.173 -0.0294 -0.479*
(0.294) (0.224) (0.254) (0.248) (0.289) (0.252) (0.254) (0.323) (0.257)
Working Part-Time (Under-

employed) -2.141*** -1.984** -0.439 -0.992 -0.00678 0.157 0.0789 -0.697 -2.194***
(0.764) (0.828) (0.765) (1.012) (0.744) (0.713) (0.767) (1.398) (0.701)
Unemployed -1.566*** -1.144*** 0.609 -1.045** 0.812** 0.265 0.290 -0.273 -0.800**
(0.408) (0.386) (0.433) (0.439) (0.395) (0.343) (0.371) (0.569) (0.400)
Out of Labour Force -0.113 0.284 -0.0572 0.169 0.0950 -0.347 0.103 0.00220 -0.121
(0.409) (0.385) (0.409) (0.397) (0.380) (0.302) (0.308) (0.433) (0.440)
Prefer not to say -0.507 -0.396 0.419 -0.369 -0.190 -0.707*** -0.400 -0.686* 0.0153
(0.388) (0.445) (0.357) (0.469) (0.306) (0.270) (0.279) (0.411) (0.343)
Income: £14,999 or Less
£15,000-£29,999 -0.0517 -0.239 0.410 -0.337 -0.0432 0.258 -0.121 -0.118 0.0925
(0.290) (0.328) (0.321) (0.320) (0.279) (0.275) (0.276) (0.405) (0.280)
£30,000-£44,999 -0.0803 -0.347 0.535 -0.126 0.0525 0.527* -0.288 -0.0209 -0.121
(0.303) (0.315) (0.329) (0.363) (0.311) (0.293) (0.306) (0.369) (0.356)
£45,000-£59,999 0.476 0.0785 0.268 -0.216 -0.106 0.100 -0.160 -0.113 0.457
(0.324) (0.365) (0.400) (0.389) (0.324) (0.316) (0.368) (0.472) (0.351)
£60,000-£74,999 0.333 0.276 0.325 -0.234 -0.237 0.0933 0.0205 -0.134 -0.0376
(0.352) (0.365) (0.369) (0.359) (0.352) (0.370) (0.350) (0.420) (0.323)
£75,000 or More 0.159 0.000426 0.0371 -0.219 -0.300 0.0649 0.0655 0.389 0.113
(0.352) (0.356) (0.342) (0.403) (0.332) (0.297) (0.367) (0.403) (0.332)
Prefer not to Say 0.834** 0.575 -0.228 0.542 -0.328 0.0827 0.268 1.195** 0.484
(0.359) (0.398) (0.363) (0.439) (0.309) (0.330) (0.415) (0.484) (0.378)
Religion: None
Christian -0.267 -0.366 -0.0735 -0.0667 -0.184 0.0929 -0.0942 -0.713*** -0.178
(0.307) (0.236) (0.243) (0.237) (0.262) (0.262) (0.199) (0.239) (0.248)
Buddhist 0.341 0.160 -0.230 0.223 -0.355 -0.191 0.538* -0.274 0.00938
(0.339) (0.324) (0.320) (0.316) (0.385) (0.337) (0.274) (0.347) (0.317)
Hindu -0.0841 -0.000969 -0.192 -0.211 -0.0174 0.291 -0.331 -0.269 -0.444
(0.555) (0.470) (0.759) (0.448) (0.519) (0.485) (0.494) (0.620) (0.557)
Jewish 0.628 -0.0508 -1.205** 0.708 -0.340 0.0246 1.695*** 1.359*** 0.530
(0.481) (0.418) (0.481) (0.507) (0.544) (0.480) (0.476) (0.482) (0.476)

Muslim -1.270** -0.831* 0.250 0.153 -0.0160 0.428 1.622*** -0.841* -3.467***
(0.560) (0.454) (0.479) (0.646) (0.532) (0.444) (0.550) (0.495) (0.551)
Sikh 0.260 -0.299 0.149 0.311 0.597 0.235 -1.125** -0.515 -1.369***
(0.582) (0.513) (0.465) (0.508) (0.706) (0.649) (0.451) (0.587) (0.459)
Other -0.449 -0.241 0.234 -0.269 -0.639 -0.203 -0.662 -1.669** -1.139
(0.344) (0.339) (0.408) (0.435) (0.397) (0.403) (0.557) (0.679) (0.824)
Prefer not to Say -0.420 -0.695* 0.827*** -0.792*** 1.357*** 1.145*** -0.0592 -0.353 -0.293
(0.317) (0.378) (0.224) (0.235) (0.333) (0.386) (0.262) (0.445) (0.325)
Religious Practice: Never
Less Than Annually 0.130 0.434* -0.0805 0.251 -0.386 -0.333 -0.296 -0.0603 -0.128
(0.291) (0.231) (0.268) (0.300) (0.257) (0.237) (0.316) (0.320) (0.321)
At Least Annually 0.00521 -0.179 0.483** 0.0319 0.290 0.182 -0.0124 0.114 0.299
(0.275) (0.230) (0.237) (0.263) (0.312) (0.279) (0.210) (0.244) (0.275)
At Least Monthly 0.154 0.202 -0.0144 0.580* -0.153 -0.129 -0.186 0.368 0.622*
(0.320) (0.304) (0.308) (0.339) (0.296) (0.348) (0.324) (0.352) (0.365)
At Least Weekly 0.230 0.147 0.177 0.128 0.209 0.0621 0.328 0.708** 0.803**
(0.366) (0.336) (0.342) (0.314) (0.333) (0.352) (0.299) (0.355) (0.321)
Prefer not to Say -0.0879 0.0410 0.979*** -0.0982 -0.0190 0.101 -0.0827 -0.379 0.286
(0.452) (0.418) (0.261) (0.382) (0.520) (0.464) (0.229) (0.424) (0.307)
Smoking: Yes
No 0.271 0.204 0.0723 0.482* -0.405 -0.262 -0.00821 0.0993 0.401

(0.224) (0.240) (0.348) (0.260) (0.263) (0.252) (0.256) (0.242) (0.277)
Pregnant: Yes
No -0.411 -0.370 -0.0892 -0.222 -0.497 -0.410 -0.252 0.221 -0.168

(0.728) (0.469) (0.609) (0.662) (0.696) (0.652) (0.362) (0.368) (0.384)
Medication: Yes
No -0.0356 0.212 -0.311* 0.134 -0.259 -0.279* -0.0975 -0.0661 -0.0421

(0.149) (0.138) (0.173) (0.157) (0.169) (0.157) (0.165) (0.169) (0.152)
Preference for Socialising:

Yes
No -0.237 -0.224 -0.0720 -0.302 0.317* 0.332* -0.396** -0.308* -0.178

(0.185) (0.168) (0.175) (0.213) (0.169) (0.180) (0.165) (0.175) (0.163)
Constant Yes Yes Yes Yes Yes Yes Yes Yes Yes
Set Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 279 279 279 279 279 279 279 279 279
R Squared 0.422 0.330 0.329 0.303 0.381 0.353 0.354 0.319 0.405
Notes: Robust standard errors clustered at individual level in parentheses. See Supplementary Materials Table 1b for variable definitions.
*** p<0.01, ** p<0.05, *
p<0.1
Table 2b
Main Results – Biomarkers (Regression Table for Figure 4), Including Controls
Stress Immune Response

Interferon Cytokine Cytokine Cytokine Chemokine
Cortisol IFN- IL-10 IL-1 IL-6 IL-8
(1) (2) (3) (4) (5) (6)
Treatment*Post 0.0248 0.207 0.0623 -0.0738 -0.0873 0.0731

(0.252) (0.246) (0.172) (0.171) (0.179) (0.200)
Treatment 0.175 -0.393* -0.0426 0.0416 0.0283 -0.158
(0.207) (0.201) (0.198) (0.192) (0.175) (0.192)
Post -0.302** -0.124 -0.149 -0.203** -0.0589 -0.288**
(0.132) (0.171) (0.112) (0.0964) (0.119) (0.123)
Age: 20-24
25-34 -0.454 0.0213 0.303 0.398 0.444 0.496

(0.486) (0.284) (0.359) (0.536) (0.297) (0.464)
35-44 -0.339 -0.437 0.180 0.236 0.318 0.409
(0.499) (0.313) (0.378) (0.560) (0.355) (0.502)
45-54 -0.602 -0.255 0.139 0.871 0.743** 0.903*
(0.496) (0.310) (0.425) (0.581) (0.335) (0.510)
55-64 -0.449 -0.167 0.160 0.955* 0.797** 1.014**
(0.548) (0.319) (0.428) (0.547) (0.365) (0.490)
65-74 -0.0713 -0.918* 0.586 0.855 0.494 0.934
(0.720) (0.502) (0.501) (0.739) (0.426) (0.662)
Gender: Male
Female -0.726*** -0.527** -0.212 -0.269 -0.476** -0.385*

(0.244) (0.245) (0.224) (0.213) (0.199) (0.224)
Partnered -0.278 -0.530** -0.0175 -0.210 -0.293 -0.181

(0.232) (0.208) (0.264) (0.254) (0.212) (0.227)
Married 0.0976 -0.417 -0.268 0.0717 0.325 -0.249
(0.479) (0.384) (0.333) (0.571) (0.411) (0.622)
Separated 0.478 -0.663** -0.292 -0.686* -0.00438 -0.287
(0.349) (0.310) (0.316) (0.412) (0.331) (0.370)
Divorced 1.871** 1.183* -0.0915 0.126 0.286 0.506
(0.849) (0.659) (0.540) (0.712) (0.636) (0.676)
Widowed -0.275 -0.0468 -0.0890 0.213 0.0978 -0.00300
(0.242) (0.231) (0.181) (0.233) (0.199) (0.236)
Prefer not to Say -0.913 -0.262 0.00989 -0.310 -0.761** 0.0821
(0.755) (0.499) (0.689) (0.533) (0.293) (0.732)
Degree
Vocational Degree -0.801* 0.781 0.429 0.596 0.591 0.863*

(0.471) (0.611) (0.350) (0.558) (0.510) (0.515)
Tertiary Degree -0.372 -0.0111 0.917*** 0.417 0.275 0.525*
(0.369) (0.372) (0.257) (0.323) (0.383) (0.315)
Higher Than Tertiary Degree -0.270 0.0780 0.627*** 0.584** 0.337 0.601**
(0.359) (0.352) (0.231) (0.273) (0.345) (0.255)
Prefer not to Say 0.0317 0.271 0.0535 -1.100 0.840 -0.103
(0.839) (0.549) (0.612) (0.730) (0.667) (0.587)
Working Full-Time for Self -0.121 0.149 0.557** 0.322 0.360 -0.173
(0.268) (0.254) (0.265) (0.321) (0.307) (0.313)
Working Part-Time -0.551* 0.439* 0.0576 -0.0869 0.0854 -0.157
(0.280) (0.255) (0.262) (0.291) (0.261) (0.276)
Working Part-Time (Underem-

ployed) 0.265 0.383 2.392** 1.383 0.0683 -0.0903
(1.231) (0.684) (1.005) (0.998) (0.855) (1.053)
Unemployed 0.479 0.823 1.200** 0.680 0.684 0.780*
(0.648) (0.594) (0.534) (0.486) (0.416) (0.456)
Out of Labour Force -0.450 0.257 0.0806 0.330 0.297 0.0438
(0.393) (0.302) (0.402) (0.361) (0.298) (0.380)
Prefer not to say -0.249 0.110 0.521 0.748* -0.121 0.0141
(0.502) (0.276) (0.432) (0.436) (0.375) (0.331)
£15,000-£29,999 0.192 0.124 0.216 -0.465 0.136 0.0999

(0.420) (0.365) (0.292) (0.391) (0.316) (0.362)
£30,000-£44,999 -0.0639 0.384 0.663** -0.000538 0.402 -0.0239
(0.400) (0.399) (0.284) (0.391) (0.338) (0.382)
£45,000-£59,999 0.135 0.320 0.812** -0.0331 0.222 0.419
(0.442) (0.440) (0.376) (0.382) (0.389) (0.390)
£60,000-£74,999 0.393 0.467 0.444 -0.163 0.280 0.253
(0.472) (0.410) (0.298) (0.346) (0.353) (0.332)
£75,000 or More -0.138 0.414 0.744** -0.107 0.356 0.307
(0.410) (0.375) (0.375) (0.387) (0.345) (0.368)
Prefer not to Say 0.311 0.297 -0.367 -0.734 -0.150 -0.0152
(0.514) (0.432) (0.478) (0.573) (0.588) (0.503)
Religion: None
Christian 0.244 0.0628 0.138 -0.0853 -0.141 -0.00697

(0.305) (0.255) (0.263) (0.286) (0.265) (0.320)
Buddhist 0.0665 0.0452 -0.223 -0.152 -0.139 -0.168
(0.420) (0.314) (0.422) (0.419) (0.475) (0.489)
Hindu 0.464 0.171 0.161 -0.321 -0.911** -0.0799
(0.509) (0.453) (0.482) (0.495) (0.437) (0.525)
Jewish -0.124 0.207 -0.163 0.271 1.124** 0.949*
(0.862) (0.521) (0.506) (0.523) (0.531) (0.563)

Muslim 0.893 1.048* 0.996** 1.790*** 0.548 1.086*
(0.706) (0.540) (0.485) (0.547) (0.525) (0.586)
Sikh 0.128 -0.963** -1.073** -1.571*** -1.294*** -0.702
(0.566) (0.477) (0.439) (0.464) (0.487) (0.481)
Other -0.0458 -0.207 -0.187 -0.0504 -0.829 -0.338
(0.449) (0.469) (0.453) (0.552) (0.523) (0.663)
Prefer not to Say 0.279 0.0557 -0.352 0.163 -0.304 0.219
(0.478) (0.347) (0.291) (0.407) (0.301) (0.267)
Less Than Annually -0.250 0.0682 -0.0493 0.0371 0.190 0.0898

(0.325) (0.317) (0.301) (0.330) (0.421) (0.368)
At Least Annually -0.470 -0.0300 0.228 0.239 -0.0887 -0.0759
(0.308) (0.249) (0.308) (0.276) (0.231) (0.301)
At Least Monthly -0.794** -0.196 0.384 0.953** 0.504 0.463
(0.350) (0.317) (0.374) (0.384) (0.355) (0.426)
At Least Weekly -0.209 0.242 -0.209 0.0878 0.465 0.120
(0.495) (0.366) (0.371) (0.428) (0.429) (0.468)
Prefer not to Say -0.539 0.229 0.632** 0.620* 0.457 0.433
(0.429) (0.286) (0.242) (0.361) (0.373) (0.331)
Smoking: Yes
No 0.144 0.226 0.225 0.211 0.193 0.0950

(0.408) (0.265) (0.222) (0.221) (0.267) (0.284)
Pregnant: Yes
No 0.549 0.458 1.163*** -0.133 0.227 0.490

(0.540) (0.371) (0.407) (0.413) (0.340) (0.611)
Medication: Yes
No -0.0400 -0.00962 -0.0461 0.0589 -0.0168 -0.124

(0.148) (0.162) (0.170) (0.175) (0.161) (0.167)

Preference for Socialising: Yes
No -0.184 0.000524 -0.235 -0.173 0.117 -0.210

(0.192) (0.192) (0.185) (0.210) (0.199) (0.212)
Constant Yes Yes Yes Yes Yes Yes

Set Fixed Effect Yes Yes Yes Yes Yes Yes
Observations 265 236 269 275 268 274
R-Squared 0.246 0.217 0.272 0.302 0.279 0.246
Notes: Robust standard errors clustered at individual level in parentheses. See Supple-
*** p<0.01, ** p<0.05, * p<0.1
Table 3a
Mechanisms – Changes in Information
Information
(a) (b) (c) (d)
Treatment*Post 0.694*** 0.651*** 0.605*** 0.549**

(0.195) (0.193) (0.183) (0.224)
Treatment 0.0878 0.0330 0.243 -0.279
(0.231) (0.206) (0.199) (0.222)
Post -0.0592 -0.0747 -0.0386 -0.0608
(0.111) (0.132) (0.128) (0.157)
Age: 20-24
25-34 0.0729 0.156 0.00778 0.0941

(0.288) (0.298) (0.275) (0.268)
35-44 -0.0832 0.0241 -0.0846 -0.0138
(0.302) (0.334) (0.322) (0.318)
45-54 -0.0166 0.178 -0.342 0.0810
(0.286) (0.331) (0.383) (0.328)
55-64 0.0555 0.275 -0.259 0.115
(0.335) (0.371) (0.328) (0.356)
65-74 -0.924* 0.119 -1.079** -0.998*
(0.483) (0.577) (0.453) (0.558)
Gender: Male
Female -0.00456 -0.0359 0.0165 0.681***

(0.261) (0.220) (0.270) (0.259)
Partnered -0.143 0.140 0.102 -0.0331

(0.185) (0.217) (0.214) (0.197)

Married 0.182 1.054*** 0.297 -0.306
(0.313) (0.315) (0.483) (0.425)
Separated -0.425 -0.0748 -0.141 -0.272
(0.354) (0.322) (0.344) (0.347)
Divorced 1.412** 1.397** 1.715*** 2.593***
(0.636) (0.647) (0.603) (0.646)
Widowed -0.275 0.148 -0.0927 -0.178
(0.236) (0.191) (0.198) (0.196)
Prefer not to Say 0.337 0.544* 0.920** 0.414
(0.497) (0.296) (0.354) (0.347)
Degree
Vocational Degree 0.0323 -0.0426 -0.417 -0.0506

(0.540) (0.544) (0.535) (0.295)
Tertiary Degree -0.100 -0.0912 -0.562** -0.414
(0.255) (0.330) (0.283) (0.257)
Higher Than Tertiary Degree -0.0892 -0.437 -0.645** -0.490**
(0.258) (0.317) (0.272) (0.223)
Prefer not to Say 0.259 -0.158 -0.691 -0.445
(0.767) (0.717) (0.599) (0.722)
Working Full-Time for Self 0.108 0.215 0.137 0.207

(0.308) (0.277) (0.292) (0.250)
Working Part-Time -0.269 -0.362 0.00442 -0.372
(0.268) (0.240) (0.262) (0.238)
Working Part-Time (Underem-
ployed) -0.915 -0.597 -1.766** -0.940
(0.778) (0.682) (0.753) (0.645)
Unemployed 0.0802 -0.457 -0.486 -0.557

(0.399) (0.424) (0.537) (0.387)
Out of Labour Force 0.256 0.337 0.425 0.0206
(0.316) (0.306) (0.399) (0.338)
Prefer not to say -0.540 -0.594 -0.193 -0.737
(0.669) (0.463) (0.376) (0.496)
£15,000-£29,999 0.0127 -0.132 0.104 -0.625**

(0.346) (0.332) (0.379) (0.303)
£30,000-£44,999 -0.0758 -0.182 -0.109 -0.631**
(0.406) (0.370) (0.411) (0.296)
£45,000-£59,999 -0.0151 0.000390 0.181 -0.758**
(0.395) (0.338) (0.392) (0.321)
£60,000-£74,999 0.0808 -0.226 0.403 -0.533*
(0.366) (0.333) (0.385) (0.319)
£75,000 or More -0.0782 -0.550 -0.00954 -0.789**
(0.459) (0.371) (0.446) (0.382)
Prefer not to Say 0.801 0.596 0.934* 0.277
(0.497) (0.410) (0.478) (0.406)
Religion: None
Christian 0.345 0.164 0.272 0.237

(0.238) (0.243) (0.236) (0.238)
Buddhist 0.443 0.000895 0.136 0.366
(0.344) (0.376) (0.315) (0.313)
Hindu -0.486 -0.775 -0.314 -0.177
(0.506) (0.524) (0.487) (0.469)
Jewish 0.459 0.677 0.386 0.794
(0.648) (0.564) (0.561) (0.502)
Muslim -0.146 0.144 0.250 -0.374
(0.713) (0.632) (0.624) (0.573)
Sikh -0.196 0.522 0.101 -0.0574

(0.569) (0.484) (0.536) (0.472)
Other -0.465 0.0813 -0.114 0.0230
(0.502) (0.399) (0.468) (0.460)
Prefer not to Say -0.466 -0.186 0.0743 0.187
(0.368) (0.353) (0.306) (0.293)
Less Than Annually -0.201 0.145 0.144 0.110

(0.372) (0.336) (0.342) (0.319)
At Least Annually -0.592* -0.237 -0.0981 -0.0421
(0.330) (0.310) (0.290) (0.274)
At Least Monthly 0.0339 0.291 0.310 0.0420
(0.291) (0.328) (0.273) (0.305)
At Least Weekly 0.165 0.602* 0.244 -0.324
(0.300) (0.317) (0.367) (0.307)
Prefer not to Say -0.405 -0.0783 -0.345 -0.00899
(0.416) (0.345) (0.418) (0.235)
Smoking: Yes
No 0.0436 0.396 0.378 0.239

(0.291) (0.284) (0.294) (0.243)
Pregnant: Yes
No -1.101** -0.687 -0.832 -0.706**

(0.497) (0.538) (0.652) (0.313)
Medication: Yes
No -0.0431 -0.243 -0.169 -0.165

(0.199) (0.196) (0.185) (0.162)
No -0.0856 -0.0995 -0.0840 -0.253

(0.175) (0.183) (0.185) (0.182)
Constant Yes Yes Yes Yes

Set Fixed Effect Yes Yes Yes Yes
Observations 230 230 230 230
R-Squared 0.357 0.328 0.316 0.352
Notes: Robust standard errors clustered at individual level in parentheses. The depend-
ent variables are (a) awareness of what contributes to a happy and meaningful life, (b)
knowledge of what really matters to oneself in life, (c) feeling of being able to do
things to improve one's own wellbeing, and (d) feeling of being able to improve the
wellbeing of others. See Supplementary Materials Table 1b for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
Table 3b
Mechanisms – Changes in Behaviour
Behaviour
(a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (l)
Treatment*Post 0.752*** 0.823*** 0.616*** 0.387* 0.282 0.430* 0.309 0.465** 0.653*** 0.409* 0.361* 0.232
(0.211) (0.226) (0.216) (0.229) (0.230) (0.252) (0.228) (0.210) (0.226) (0.210) (0.206) (0.185)
Treatment -0.152 -0.157 -0.100 0.0180 -0.267 -0.0341 -0.153 -0.265 -0.665** -0.154 -0.300 -0.0330
(0.218) (0.246) (0.207) (0.225) (0.202) (0.258) (0.204) (0.224) (0.265) (0.220) (0.192) (0.210)
Post -0.0894 -0.209* 0.0345 0.0274 0.137 0.0947 -0.172 0.0122 0.0365 -0.0116 0.380*** 0.119
(0.126) (0.126) (0.130) (0.169) (0.158) (0.171) (0.120) (0.134) (0.125) (0.141) (0.142) (0.122)
Age: 20-24
25-34 0.0605 0.840** -0.142 0.0523 0.0416 -0.0401 -0.0515 0.612 0.213 0.391 0.953*** 0.460
(0.377) (0.325) (0.341) (0.373) (0.278) (0.521) (0.365) (0.398) (0.339) (0.257) (0.297) (0.335)
35-44 0.143 0.725* -0.203 -0.0407 -0.544* -0.375 -0.351 0.276 0.242 0.461 1.172*** 0.270
(0.400) (0.432) (0.385) (0.399) (0.326) (0.530) (0.372) (0.426) (0.365) (0.321) (0.344) (0.381)
45-54 0.241 0.817* -0.332 -0.196 -0.0590 -0.290 -0.232 0.172 0.412 0.367 1.353*** 0.606
(0.411) (0.453) (0.407) (0.408) (0.348) (0.582) (0.397) (0.456) (0.362) (0.317) (0.356) (0.369)
55-64 -0.0673 0.798* 0.0693 -0.200 -0.463 -0.562 -0.421 0.521 0.440 0.241 0.921** 0.187
(0.418) (0.478) (0.453) (0.433) (0.359) (0.590) (0.401) (0.455) (0.428) (0.377) (0.386) (0.392)
65-74 -1.309** -0.153 -0.446 0.126 -0.618 -0.422 -1.589*** 0.0126 -0.315 0.580 1.234** 0.428
(0.548) (0.512) (0.648) (0.694) (0.524) (0.642) (0.535) (0.572) (0.505) (0.737) (0.565) (0.481)
Gender: Male
Female 0.625** 0.159 0.210 -0.0320 0.339 0.124 0.220 0.473* 0.869*** 0.535** 0.805*** 0.489**
(0.253) (0.291) (0.237) (0.251) (0.221) (0.288) (0.263) (0.277) (0.253) (0.250) (0.213) (0.215)
Partnered 0.0707 0.173 -0.0316 0.412* 0.369** 0.318 0.717*** 0.285 0.0894 -0.237 0.180 -0.140
(0.198) (0.251) (0.225) (0.217) (0.183) (0.239) (0.225) (0.251) (0.228) (0.227) (0.192) (0.184)
Married 0.243 -0.00998 0.870 0.367 0.198 0.155 0.123 0.0803 0.238 0.814* -0.406 -0.0452
(0.490) (0.488) (0.546) (0.469) (0.473) (0.501) (0.479) (0.608) (0.687) (0.482) (0.462) (0.380)
Separated -0.0316 0.268 0.0684 0.207 0.0177 0.0919 -0.113 -0.199 -0.0403 -0.121 0.335 0.0212
(0.430) (0.396) (0.383) (0.405) (0.393) (0.424) (0.292) (0.366) (0.516) (0.364) (0.378) (0.417)
Divorced 0.648 1.563** -0.321 -0.390 -0.405 -1.749*** 2.955*** 1.334** -0.173 -0.728 0.899 0.561
(0.543) (0.676) (0.736) (0.763) (0.630) (0.617) (0.621) (0.605) (0.681) (0.848) (0.604) (0.604)
Widowed -0.0707 0.274 0.0231 0.218 0.166 0.198 0.323 0.103 0.139 -0.0987 -0.0504 -0.0780
(0.228) (0.267) (0.201) (0.221) (0.185) (0.255) (0.223) (0.214) (0.244) (0.197) (0.205) (0.221)
Prefer not to Say 1.095** -0.478 1.376** 0.853** -0.944* 1.625* -0.627* -0.851** -0.425 -0.287 0.739*** -0.437
(0.498) (0.802) (0.551) (0.386) (0.494) (0.914) (0.376) (0.427) (0.375) (0.330) (0.265) (0.399)
Educational Status: Secondary Degree
Vocational Degree -0.445 0.262 -1.038** -0.962 -0.449 -0.404 0.550 -0.905* -0.290 0.226 -0.385 -0.349
(0.568) (0.404) (0.523) (0.621) (0.465) (0.469) (0.375) (0.521) (0.466) (0.510) (0.415) (0.586)
Tertiary Degree -0.407 0.219 -1.002*** -0.663* -0.630* -0.115 0.465 -0.598 -0.407 0.470 -0.109 -0.277
(0.425) (0.375) (0.364) (0.388) (0.347) (0.368) (0.329) (0.373) (0.291) (0.344) (0.217) (0.417)
Higher Than Tertiary Degree -0.599 0.0252 -1.196*** -0.871** -0.561* -0.310 0.153 -0.935** -0.868*** 0.0449 -0.335 -0.0122
(0.405) (0.354) (0.327) (0.364) (0.314) (0.338) (0.324) (0.385) (0.275) (0.326) (0.208) (0.400)
Prefer not to Say -0.488 -1.570* 0.139 -0.637 -1.980*** 0.249 1.329** 0.0421 0.800 -0.552 -0.266 -0.284
(0.681) (0.836) (0.603) (0.708) (0.593) (0.633) (0.582) (0.665) (0.785) (0.626) (0.515) (0.690)
Employment Status: Working Full-Time
for Employer
Working Full-Time for Self 0.317 -0.122 0.291 0.218 0.470* 0.127 -0.0472 0.146 0.149 0.499* 0.526* 0.338
(0.315) (0.306) (0.291) (0.289) (0.245) (0.317) (0.299) (0.298) (0.362) (0.293) (0.295) (0.307)
Working Part-Time 0.103 -0.215 -0.0963 0.0229 -0.129 0.197 -0.0404 -0.381 0.0183 -0.118 -0.329 0.188
(0.270) (0.315) (0.253) (0.274) (0.222) (0.323) (0.241) (0.275) (0.319) (0.285) (0.231) (0.259)
Working Part-Time (Underemployed) -0.721 1.675 -4.005*** -2.464*** -0.0200 -3.579*** -1.730** -0.0992 0.579 0.00843 0.476 -0.428
(0.871) (1.117) (0.884) (0.862) (0.769) (1.182) (0.743) (0.871) (0.856) (0.731) (0.684) (0.771)
Unemployed -0.633 -0.0557 -0.181 -0.242 0.195 0.513 0.0428 -0.276 0.283 -0.955** 0.315 0.212
(0.384) (0.557) (0.405) (0.395) (0.439) (0.486) (0.389) (0.461) (0.389) (0.369) (0.448) (0.341)
Out of Labour Force 0.863* 0.0516 0.783 0.500 0.242 0.610 0.582 0.212 0.419 0.126 0.328 0.670
(0.484) (0.437) (0.511) (0.487) (0.513) (0.515) (0.363) (0.413) (0.442) (0.542) (0.531) (0.433)
Prefer not to say 0.00643 0.283 -0.0814 0.339 -0.527* -0.500 0.149 -1.303*** -0.526 -0.506 -0.417 -0.540
(0.352) (0.542) (0.378) (0.384) (0.294) (0.381) (0.266) (0.373) (0.456) (0.378) (0.425) (0.399)
£15,000-£29,999 -0.241 -0.0169 -0.364 -0.142 -0.353 0.0478 -0.366 0.0143 -0.00364 -0.149 -0.175 -0.195
(0.410) (0.443) (0.380) (0.391) (0.353) (0.388) (0.363) (0.357) (0.343) (0.327) (0.314) (0.375)
£30,000-£44,999 -0.0761 0.161 -0.0125 -0.131 0.241 -0.235 -0.578* -0.197 -0.0511 -0.0348 -0.557* 0.0248
(0.372) (0.396) (0.413) (0.418) (0.325) (0.400) (0.308) (0.365) (0.391) (0.377) (0.308) (0.358)
£45,000-£59,999 0.158 0.00964 0.209 0.340 0.285 0.173 0.192 0.371 0.205 0.393 -0.170 -0.0240
(0.405) (0.453) (0.428) (0.404) (0.370) (0.424) (0.391) (0.463) (0.420) (0.423) (0.362) (0.372)
£60,000-£74,999 -0.0111 -0.239 0.407 0.224 0.163 -0.0547 0.320 0.0509 0.195 0.0671 -0.668* -0.238
(0.398) (0.459) (0.443) (0.407) (0.384) (0.427) (0.363) (0.446) (0.456) (0.368) (0.377) (0.361)
£75,000 or More -0.0332 0.0728 0.162 -0.0191 0.537 0.413 -0.342 -0.236 -0.0108 -0.209 -0.307 0.142
(0.410) (0.488) (0.455) (0.424) (0.353) (0.450) (0.414) (0.415) (0.420) (0.416) (0.358) (0.356)
Prefer not to Say 0.549 0.0841 0.956** 0.838* 1.145*** 0.452 0.155 1.186*** 0.474 0.300 -0.0796 0.381
(0.434) (0.547) (0.482) (0.462) (0.377) (0.422) (0.377) (0.413) (0.531) (0.495) (0.441) (0.493)
Religion: None
Christian 0.103 -0.269 0.235 0.0498 0.0838 -0.241 -0.130 -0.451* 0.384 0.345 -0.0223 -0.356
(0.265) (0.379) (0.278) (0.270) (0.243) (0.287) (0.228) (0.253) (0.299) (0.217) (0.233) (0.261)
Buddhist 0.112 0.0917 0.594* 0.374 0.696* -0.276 0.378 0.0405 0.783* 0.101 -0.0270 -0.128
(0.390) (0.460) (0.302) (0.397) (0.357) (0.369) (0.338) (0.422) (0.414) (0.389) (0.355) (0.377)
Hindu -0.453 -0.784 -0.186 0.00196 0.229 0.163 0.685 0.112 -0.0664 0.298 -0.449 -0.383
(0.624) (0.698) (0.682) (0.660) (0.726) (0.616) (0.563) (0.595) (0.811) (0.660) (0.516) (0.539)
Jewish 0.796 0.301 -0.790 -0.517 -0.498 -0.0696 -1.648*** -1.546*** -0.606 0.0573 0.581 1.073**
(0.535) (0.621) (0.548) (0.495) (0.506) (0.549) (0.504) (0.531) (0.566) (0.592) (0.555) (0.522)
Muslim 0.391 0.526 0.777 0.442 0.662 0.205 0.717 0.486 2.033*** -0.0307 0.0806 -0.550
(0.668) (0.694) (0.564) (0.552) (0.533) (0.644) (0.534) (0.555) (0.564) (0.597) (0.545) (0.567)
Sikh -1.409** -0.150 0.423 0.449 0.395 0.685 -0.232 0.182 0.306 0.172 1.073** -0.181
(0.684) (0.539) (0.432) (0.661) (0.484) (0.594) (0.508) (0.488) (0.660) (0.503) (0.435) (0.586)
Other 0.0638 -0.387 0.0783 -0.0547 -0.547 -0.313 -0.693** -1.202** 0.125 0.0788 -0.786** -0.0784
(0.534) (0.524) (0.403) (0.451) (0.406) (0.443) (0.298) (0.588) (0.567) (0.403) (0.360) (0.413)
Prefer not to Say -0.0936 -0.652* 0.0356 -0.138 -0.0559 -0.546** -0.268 0.0207 0.00176 0.522** -0.191 0.382
(0.384) (0.332) (0.222) (0.407) (0.330) (0.265) (0.321) (0.307) (0.363) (0.263) (0.335) (0.314)
Less Than Annually -0.156 0.433 -0.0293 -0.352 -0.412 0.0686 -0.318 -0.0204 -0.0182 -0.0536 0.197 0.128
(0.365) (0.420) (0.342) (0.310) (0.305) (0.384) (0.297) (0.315) (0.385) (0.307) (0.339) (0.446)
At Least Annually 0.108 0.00657 -0.195 -0.211 0.0345 0.0440 -0.152 0.00826 -0.185 -0.0167 0.123 0.0115
(0.317) (0.331) (0.253) (0.283) (0.239) (0.307) (0.237) (0.282) (0.291) (0.264) (0.258) (0.264)
At Least Monthly 0.576 1.108*** 0.457 -0.0822 0.0512 0.544 -0.389 0.0615 -0.382 0.174 0.467* 0.490*
(0.404) (0.398) (0.313) (0.370) (0.308) (0.332) (0.298) (0.384) (0.374) (0.323) (0.272) (0.289)
At Least Weekly 0.647* 0.883 0.117 0.371 0.246 1.007** 0.265 0.559 0.0327 0.182 0.322 0.565
(0.337) (0.535) (0.372) (0.359) (0.381) (0.403) (0.316) (0.392) (0.368) (0.331) (0.346) (0.410)
Prefer not to Say 0.323 0.996*** -0.149 -0.0530 0.207 0.660* 0.0938 -0.165 0.160 0.526 0.941*** 0.801*
(0.473) (0.325) (0.229) (0.476) (0.271) (0.351) (0.329) (0.317) (0.387) (0.331) (0.268) (0.408)
Smoking: Yes
No 0.732* 0.211 0.749** 0.491 -0.234 -0.211 0.645** 0.310 0.698** -0.0389 -0.0755 -0.238
(0.414) (0.393) (0.363) (0.317) (0.385) (0.411) (0.293) (0.348) (0.301) (0.319) (0.353) (0.333)
Pregnant: Yes
No -0.548 -0.262 -0.852 0.263 0.637 0.919* 0.376 0.0602 0.230 -0.230 0.205 -0.356
(0.832) (0.385) (0.727) (0.960) (0.798) (0.508) (0.898) (0.699) (0.698) (0.327) (0.688) (0.665)
Medication: Yes
No -0.0356 -0.00986 -0.0374 0.0517 0.193 0.116 0.137 -0.103 0.217 0.132 0.0377 0.203
(0.179) (0.203) (0.179) (0.180) (0.146) (0.199) (0.174) (0.188) (0.213) (0.189) (0.141) (0.178)
No -0.0740 0.147 0.166 -0.0932 -0.287 -0.194 0.181 0.0495 0.0574 -0.293* -0.343** 0.0298
(0.200) (0.238) (0.194) (0.213) (0.174) (0.215) (0.216) (0.235) (0.221) (0.167) (0.148) (0.181)
Constant Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Set Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 230 230 230 230 230 230 230 230 230 230 230 230
R Squared 0.336 0.316 0.376 0.263 0.339 0.294 0.399 0.313 0.307 0.298 0.442 0.306
Notes: Robust standard errors clustered at individual level in parentheses. The dependent variables are the frequency in recent weeks of (a) noticing and feeling grateful for good things, (b) practising mindfulness
or meditation, (c) treating oneself in a kind way, (d) making time for something really important for oneself, (e) responding well to difficult situations, (f) learning or trying out something new, (g) giving time to
one of oneself's closest relationships, (h) connecting with other people, (i) doing something kind or helpful for others, (j) trying to increase happiness at work, (k) trying to increase happiness in the community,
and (l) thinking about the difference one makes to the world. See Supplementary Materials Table 1b for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
Table 4
Absence of Time Trend for Control Group
Mean Mean Difference

Control Group, Control Group,
Pre-Intervention Post-Intervention
Life Satisfaction 6.315 6.271 0.044

Happiness 6.288 5.957 0.331
Anxiousness 4.438 4.500 -0.062
Worthwhileness 7.041 6.971 0.070
PHQ-9 (Depression) 7.151 7.057 0.094
GAD-7 (Anxiety) 6.671 7.086 -0.414
Compassion 6.792 6.657 0.135
Social Trust 6.507 6.386 0.121
Gratitude 6.178 6.114 0.064
Information (a) 7.469 7.271 0.198
Information (b) 7.224 6.943 0.282
Information (c) 6.796 6.629 0.167
Information (d) 7.224 7.171 0.053
Behaviour (a) 2.020 1.829 0.192*
Behaviour (b) 1.429 1.129 0.300***
Behaviour (c) 1.449 1.471 -0.022
Behaviour (d) 1.571 1.514 0.057
Behaviour (e) 1.571 1.586 -0.014
Behaviour (f) 1.204 1.129 0.076
Behaviour (g) 1.837 1.629 0.208*
Behaviour (h) 2.061 1.929 0.133
Behaviour (i) 1.816 1.786 0.031
Behaviour (j) 1.347 1.314 0.033

Behaviour (k) 0.776 0.986 -0.210*
Behaviour (l) 1.122 1.214 -0.092
Biomarkers
Cortisol 0.172 0.102 0.070

Interferon IFN- 7.510 7.733 -0.223
Cytokine IL-10 1.397 1.433 -0.037
Cytokine IL-1 284.884 214.072 70.813***
Cytokine IL-6 10.131 8.783 1.348
Chemokine IL-8 1,603.446 1,287.056 316.390***
Controls
Age: 20-24 0.068 0.057 0.011

25-34 0.247 0.243 0.004
35-44 0.233 0.229 0.004
45-54 0.205 0.214 -0.009*
55-64 0.192 0.200 -0.008*
65-74 0.055 0.057 -0.002
Gender: Male 0.164 0.171 -0.007
Female 0.836 0.829 0.007
Married 0.192 0.200 -0.008
Separated 0.041 0.043 -0.002
Divorced 0.096 0.100 -0.004
Widowed 0.014 0.014 -0.001
Domestic Partner 0.219 0.200 0.019
Educational Status: Secondary Degree 0.041 0.043 -0.002
Vocational Degree 0.068 0.071 -0.003
Tertiary Degree 0.356 0.357 -0.001

Higher Than Tertiary Degree 0.534 0.529 0.006
Employment Status: Working Full-Time 0.466 0.457 0.009
for Employer
Working Full-Time for Self 0.164 0.171 -0.007
Working Part-Time 0.219 0.229 -0.009
Working Part-Time (Underemployed) 0.000 0.000 0.000
Unemployed 0.041 0.043 -0.002
Out of Labour Force 0.096 0.086 0.010
Prefer not to say 0.014 0.014 -0.001
Income: £14,999 or Less 0.137 0.143 -0.006
£15,000-£29,999 0.192 0.186 0.006
£30,000-£44,999 0.233 0.229 0.004
£45,000-£59,999 0.178 0.171 0.007
£60,000-£74,999 0.110 0.114 -0.005*
£75,000 or More 0.123 0.129 -0.005*
Prefer not to Say 0.027 0.029 -0.001
Religion: None 0.589 0.586 0.003
Christian 0.247 0.243 0.004
Buddhist 0.055 0.057 -0.002
Hindu 0.014 0.014 -0.001
Jewish 0.000 0.000 0.000
Muslim 0.000 0.000 0.000
Sikh 0.000 0.000 0.000
Other 0.041 0.043 -0.002
Prefer not to Say 0.055 0.057 -0.002
Religious Practice: Never 0.521 0.529 -0.008
Less Than Annually 0.110 0.100 0.010
At Least Annually 0.123 0.114 0.009
At Least Monthly 0.110 0.114 -0.005
At Least Weekly 0.096 0.100 -0.004
Prefer not to Say 0.041 0.043 -0.002

Smoking: Yes 0.110 0.086 0.024
No 0.890 0.914 -0.024
Pregnant: Yes 0.027 0.029 -0.001
No 0.973 0.971 0.001
Medication: Yes 0.438 0.429 0.010
No 0.562 0.571 -0.010
Importance of Meeting New People and 0.808 0.814 -0.006
Making Friends: Yes
No 0.192 0.186 0.006
Observations 73 70 -
Notes: T-tests used robust standard errors clustered at individual level. See Supple-
*** p<0.01, ** p<0.05, * p<0.1
Table 5a
Main Results – Self-Reported Outcomes (Regression Table for Figure 3), Adjusted for Multiple Hypotheses Testing (Stepdown P-Values)
Subjective Wellbeing Mental Health Pro-Sociality

PHQ-9
Life Satis- Anxious- Worth- (Depres- GAD-7
faction Happiness ness whileness sion) (Anxiety) Compassion Social Trust Gratitude
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Treatment*Post 0.645 0.631 -0.425 0.565 -0.543 -0.448 0.422 0.561 0.278
(0.144) (0.157) (0.174) (0.146) (0.116) (0.114) (0.127) (0.148) (0.125)
P Value (Rounded) 0.000 (***) 0.000 (***) 0.016 (**) 0.000 (***) 0.000 (***) 0.000 (***) 0.001 (***) 0.000 (***) 0.028 (**)
Stepdown P Value (Rounded) 0.010 (***) 0.010 (***) 0.059 (*) 0.010 (***) 0.010 (***) 0.010 (***) 0.010 (***) 0.010 (***) 0.059 (*)
Treatment 0.062 -0.020 -0.111 -0.071 -0.083 -0.141 -0.192 -0.126 0.014
(0.166) (0.146) (0.166) (0.157) (0.157) (0.152) (0.159) (0.177) (0.163)
Post -0.024 -0.143 0.027 -0.046 -0.024 0.050 -0.101 -0.068 -0.065
(0.095) (0.116) (0.127) (0.100) (0.080) (0.088) (0.086) (0.092) (0.083)
Constant Yes Yes Yes Yes Yes Yes Yes Yes Yes
Controls No No No No No No No No No
Set Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 279 279 279 279 279 279 279 279 279
R Squared 0.107 0.081 0.061 0.062 0.099 0.069 0.026 0.033 0.030
Notes: Robust standard errors clustered at individual level in parentheses. See Supplementary Materials Table 1b for variable definitions.
*** p<0.01, ** p<0.05, * p<0.1
Table 5b
Main Results – Biomarkers (Regression Table for Figure 4), Adjusted for Multiple Hypotheses Testing (Stepdown P-Values)
Stress Immune Response

Interferon Cytokine Cytokine Cytokine Chemokine
Cortisol IFN- IL-10 IL-1 IL-6 IL-8
(1) (2) (3) (4) (5) (6)
Treatment*Post -0.007 0.281 0.045 -0.049 -0.047 0.080

(0.235) (0.223) (0.162) (0.149) (0.163) (0.183)
P Value (Rounded) 0.976 0.211 0.780 0.745 0.773 0.662
Stepdown P Value (Rounded) 0.990 0.990 0.990 0.990 0.990 0.980
Treatment 0.150 -0.395** -0.019 0.001 -0.047 -0.123

(0.208) (0.171) (0.184) (0.175) (0.169) (0.172)
Post -0.255** -0.131 -0.146 -0.182** -0.057 -0.261**
(0.121) (0.154) (0.104) (0.087) (0.112) (0.116)
Constant Yes Yes Yes Yes Yes Yes

Controls No No No No No No
Set Fixed Effect Yes Yes Yes Yes Yes Yes
Observations 265 236 269 275 268 274
R-Squared 0.021 0.028 0.007 0.016 0.012 0.018
Notes: Robust standard errors clustered at individual level in parentheses. See Supple-
*** p<0.01, ** p<0.05, * p<0.1
Supplementary Materials II:
Course Materials
The recruitment process of course leaders is documented at:
http://www.actionforhappiness.org/media/498423/exploring_what_matters_course_leader.pdf
The course materials for course participants can be found at:
http://www.actionforhappiness.org/media/508643/exploring_what_matters.pdf
The course materials for course leaders can be found at:
http://www.actionforhappiness.org/media/519959/course_leader_guide.pdf
Supplementary Materials III:
Impact Evaluation Materials
Table of Contents
1. Project Information Sheet
2. Written Consent Form
3. Baseline Survey
4. Endline Survey
5. Follow-up Survey
EXPLORING WHAT MATTERS – PROJECT INFORMATION
You are being invited to take part in a research project. Before you decide, it is important for you
to understand why the research is being done and what it will involve. Please take time to read
the following information and ask us if there is anything that is not clear. Thank you
PROJECT OVERVIEW
This study aims to learn about the psychological and physiological wellbeing of people taking part in the
E ploring What Matters ourse de eloped A tion for Happiness. Parti ipation is oluntar and ou ha e
the right to withdraw at any point without needing to give any reason.
YOUR INVOLVEMENT
Taking part in this study involves attending the 8-week Exploring What Matters ourse, which has been run
successfully many times in local communities across the UK. On three occasions - before, during and after the
course - you will also be invited to provide some information about your personal wellbeing and attitudes, plus
a small salivary sample to allow the measurement of biomarkers.
WHAT ARE BIOMARKERS?

Biomarkers are measurable features that can be used to predict physiological states. For this study, salivary
samples will be used to measure specific hormones and proteins relating to wellbeing, with a particular focus
on stress response and immune function. No markers for any other illnesses will be investigated. The
biomarker samples may be shipped to a laboratory outside the UK for analysis.
DATA CONFIDENTIALITY
All data collected as part of this project will be treated confidentially. You will be given an identification
number so that all the data you provide will be anonymous and you cannot be identified by it. Any data that is
retained will be kept securely in accordance with the Data Protection Act.
ARE THERE ANY RISKS INVOLVED?

We do not envisage any risks from taking part in this project. If you don t enjo the ourse or don t ish to
provide the required information or salivary samples, you can withdraw at any time. While it is unlikely, it is
possible you may experience distress at some point as the study asks you to consider your emotions. If this
occurs you may wish to seek advice from your GP or a counsellor, therapist or other professional.
PROJECT OUTCOMES
The results of the project may be published in academic journals and books as well as in other forms (e.g.
reports, websites) in the public domain. You will not be identified by name or other identifying feature in any
publication. If you are interested, we can send these results to you once the project is complete.
WHO IS BEHIND THE STUDY?

The study is being led by the Centre for Economic Performance at the London School of Economics, with
funding from the Templeton Foundation. It is being run in collaboration with Action for Happiness.
THANK YOU
We are very grateful for your participation in this study. Your responses will help to provide valuable insight
into the wellbeing of individuals who participate in the Exploring What Matters course.
For more information about the project, please contact:

LSE / CEP: Dr Jan-Emmanuel de Neve 020 7955 7447 j.de-neve@lse.ac.uk
Action for Happiness: Dr Mark Williamson 020 8980 6263 mark.williamson@actionforhappiness.org
EXPLORING WHAT MATTERS – CONSENT FORM
I have read the Information Sheet relating to this research study and have been provided
with the opportunity to discuss any details or questions about this.
I understand the aims of this research and the procedures which I will be involved with as
part of the study, including providing salivary samples and information about my wellbeing.
I understand that all data relating my involvement in this study will remain confidential and
the researchers involved will not be able to identify me by my responses as my data are
anonymous. I also understand that the samples provided may be shipped to a laboratory
outside the UK for analysis. I have been informed what will happen when the study has
been completed.
I fully and freely provide my consent to participate in this study. By giving this consent, I also
understand that at any time during the study, I have the right to withdraw without
disadvantage to myself and I will not be required to provide a reason. I can withdraw from
the study by sending an email to info@actionforhappiness.org.
I also understand that if I do decide to withdraw, the researchers reserve the right to use
the anonymous data which I have provided when analysing and writing-up the study.
Name of participant (please print): ________________________________________
Date: ________________________________________
Signature: ________________________________________
EXPLORING WHAT MATTERS – PARTICIPANT QUESTIONS
SECTION I: YOUR WELLBEING
Please answer these 4 questions o a s ale of to where ea s Not at all a d ea s Co pletely .
Overall, how satisfied are you with your life nowadays?

Not at all Completely
0 1 2 3 4 5 6 7 8 9 10
Overall, how happy did you feel yesterday?

0 1 2 3 4 5 6 7 8 9 10
Overall, how anxious did you feel yesterday?

0 1 2 3 4 5 6 7 8 9 10
Overall, to what extent do you feel the things you do in your life are worthwhile?
0 1 2 3 4 5 6 7 8 9 10
Over the last two weeks, how often have you been More Nearly
Several
Not at all than half every
bothered by any of the following problems? days
the days day
1. Little interest or pleasure in doing things 0 1 2 3
2. Feeling down, depressed or hopeless 0 1 2 3
3. Trouble falling or staying asleep, or sleeping too much 0 1 2 3
4. Feeling tired or having little energy 0 1 2 3
5. Poor appetite or overeating 0 1 2 3
6. Feeling bad about yourself – or that you are a failure or 0 1 2 3

have let yourself or your family down
7. Trouble concentrating on things, such as reading the 0 1 2 3

newspaper or watching television
8. Moving or speaking so slowly that other people could 0 1 2 3

have noticed? Or the opposite – being so fidgety or
restless that you have been moving a lot more than usual
9. Thoughts that you would be better off dead or hurting 0 1 2 3

yourself in some way
Page 1 of 7
Several
bothered by the following problems? days
the days day
1. Feeling nervous, anxious or on edge 0 1 2 3
2. Not being able to stop or control worrying 0 1 2 3
3. Worrying too much about different things 0 1 2 3
4. Trouble relaxing 0 1 2 3
5. Being so restless that it is hard to sit still 0 1 2 3
6. Becoming easily annoyed or irritable 0 1 2 3
7. Feeling afraid as if something awful might happen 0 1 2 3
SECTION II: YOUR ATTITUDES
COMPASSION
Please answer the following questions honestly and quickly using the scale below.
When I hear about someone (a stranger) going through a difficult time, I feel a great deal of
compassion for him or her
Not at all true of me Very true of me
1 2 3 4 5 6 7
I tend to feel compassion for people, even though I do not know them
1 2 3 4 5 6 7
One of the activities that provides me with the most meaning to my life is helping others in the
world when they need help
1 2 3 4 5 6 7
I would rather engage in actions that help others, even though they are strangers, than engage in
actions that would help me
1 2 3 4 5 6 7
I often have tender feelings toward people (strangers) when they seem to be in need
1 2 3 4 5 6 7
Page 2 of 7
TRUST
Ge erally speaki g, would you say that ost people a e trusted, or that you a ’t e too
careful in dealing with people?
Ca ’t e too areful Most people can be trusted

0 1 2 3 4 5 6 7 8 9 10
GRATITUDE
I have so much in life to be thankful for
Strongly Slightly Strongly

Disagree Neutral Slightly Agree Agree
Disagree Disagree Agree
1 2 3 4 5 6 7
SOCIAL CONTEXT
How important is it for you to be meeting new people and making friends?
Not at all important Extremely important

0 1 2 3 4 5 6 7 8 9 10
How regularly do you meet with local groups (e.g. club, residents association, choir etc)?
Never Extremely often

0 1 2 3 4 5 6 7 8 9 10
SECTION III: ABOUT YOU
Please circle the answers which apply:
What age are you? 16-19 20-24 25-34 35-44 45-54 55-64 65-74 75+
What is your gender? Male Female
Page 3 of 7
Tick here
What is your Less than £15,000
approximate annual £15,000 to £29,999
household income?
£30,000 to £44,999
£45,000 to £59,999
£60,000 to £74,999
£75,000 or more
Prefer not to say
What is your current Single/never been married

marital status? Married
Separated
Divorced
Widowed
Domestic partner
Prefer not to say
What is your Employed full time for an employer

employment status? Employed full time for self
Employed part time - do not want full time
Employed part time - want full time
Unemployed
Out of workforce
Prefer not to say
What is your highest Completed elementary education or less

completed level of Secondary education
education?
Tertiary education – vocational
Tertiary education – graduate
Tertiary education – post-graduate
Prefer not to say
Page 4 of 7
Tick here
What is your ethnic White

group? Indian
Pakistani
Bangladeshi
Chinese
African
Caribbean
Arab
Mixed
Other
Prefer not to say
Do you regard yourself No religion

as belonging to any Christian (including Church of England, Catholic, Protestant)
particular religion?
Buddhist
If yes, which?
Hindu
Jewish
Muslim
Sikh
Other religion (please specify):
Prefer not to say
Apart from such special Never

occasions as weddings, Less than annually
funerals and baptisms,
At least annually
how often nowadays do
you attend services or At least monthly
meetings connected At least weekly
with your religion?
Prefer not to say
Page 5 of 7
How much do you care about eating a healthy diet?
Not at all Very Much

0 1 2 3 4 5 6 7 8 9 10
How often do you drink alcohol?
Not at all Very Often

0 1 2 3 4 5 6 7 8 9 10
How often are you physically active (for example, by doing sports)?
Not at all Very Often

0 1 2 3 4 5 6 7 8 9 10
How many hours did you sleep last night?
How satisfied are you with your sleep last night?

0 1 2 3 4 5 6 7 8 9 10
Do you smoke? Yes No
Are you pregnant? Yes No
Are you currently on Yes No

any medication?
If yes please specify:
Page 6 of 7
SECTION III: YOUR LIFE
Please respond to the following statements honestly and quickly using the scale below.
I feel aware of what contributes to a happy and meaningful life

0 1 2 3 4 5 6 7 8 9 10
I know what really matters to me in life

0 1 2 3 4 5 6 7 8 9 10
I feel able to do things to improve my own wellbeing

0 1 2 3 4 5 6 7 8 9 10
I feel able to do things to improve the wellbeing of others

0 1 2 3 4 5 6 7 8 9 10
In recent weeks, how often have you done the More Nearly
Several
following? days
the days day
1. Noticed and felt grateful for good things 0 1 2 3
2. Practised mindfulness / meditation 0 1 2 3
3. Treated yourself in a kind way 0 1 2 3
4. Made time for something really important to you 0 1 2 3
5. Responded well to a difficult situation 0 1 2 3
6. Learnt or tried out something new 0 1 2 3
7. Gave time to one of your closest relationships 0 1 2 3
8. Connected with other people 0 1 2 3
9. Did something kind or helpful for others 0 1 2 3
10. Tried to increase happiness at work 0 1 2 3
11. Tried to increase happiness in the community 0 1 2 3
12. Thought about the difference you make in the world 0 1 2 3
Many thanks for taking the time to provide this information.

Please check to make sure that you have answered all the questions.
Page 7 of 7

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10
Several
the days day




Page 1 of 5
Several
the days day
COMPASSION
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
Page 2 of 5
TRUST

0 1 2 3 4 5 6 7 8 9 10
GRATITUDE

1 2 3 4 5 6 7
SOCIAL CONTEXT

0 1 2 3 4 5 6 7 8 9 10
HEALTH

any medication?
Page 3 of 5

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10
Several
following? days
the days day
Page 4 of 5
SECTION IV: YOUR COURSE
How many of the 8 Exploring What Matters course sessions did you attend?
None All
0 1 2 3 4 5 6 7 8
How was the course overall?

1 2 3 4 5
Very poor Poor Ok Good Very good
How was the content of the sessions?

1 2 3 4 5
How was the facilitation of the course?

1 2 3 4 5
How likely are you to recommend the course to others?

Not at all likely Extremely likely
0 1 2 3 4 5 6 7 8 9 10
What were the highlights of the course for you?
What could we do to make the course even better?
How would you describe the impact of the course on your life?
Negative Neutral Positive
What will you do differently as a result of the course?

Page 5 of 5

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10
Several
the days day




Page 1 of 4
Several
the days day
COMPASSION
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
Page 2 of 4
TRUST

0 1 2 3 4 5 6 7 8 9 10
GRATITUDE

1 2 3 4 5 6 7
SOCIAL CONTEXT

0 1 2 3 4 5 6 7 8 9 10
HEALTH

any medication?
Page 3 of 4

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10
Several
following? days
the days day

Page 4 of 4

DP 14447

Uploaded by

Copyright:

Available Formats

DP 14447

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DP 14447

Uploaded by

Copyright:

Available Formats

DISCUSSION PAPER SERIES

IZA DP No. 14447

A Local Community Course That Raises

IZA DP No. 14447

A Local Community Course That Raises

Christian Krekel Daisy Fancourt

IZA – Institute of Labor Economics

JEL Classification: C93, I12, I31

ance with lockdown measures during Covid-19 (Krekel et al., 2020).

may be limited in societies with already high economic living standards.

Another point of view, in line with expectancy-value theory in psychology (Battle,

terventionist view, wellbeing is malleable and significant increases in average population

wellbeing may be possible, even in economically affluent societies.

tended following-up period.

an outcome for public policy more generally.

want to share these ideas with others.

The intervention is manualised: each course is led by two volunteers as facilitators on

an unpaid basis.2 Recruitment of course leaders follows a documented, standardised process:

plete documentation of the recruitment process of course leaders.

questions is discussed against the background of scientific evidence on subjective wellbeing,

mental health, and pro-sociality as well as motivation and group learning.

Hypotheses. Course design and delivery are rooted in psychological self-determina-

course materials of both course participants and course leaders.

There is an established evidence base linking psychological self-determination theory

particular mindfulness, meditation, and related self-regulation strategies, to positive outcomes

in non-clinical populations, including wellbeing, depression, and anxiety, with medium to

adaptation – the course may have sustained impacts.

ple from the nationally representative UK Household Longitudinal Survey ("Understanding

about 47% of a standard deviation), pre-treatment, than the general population.8

Randomised Controlled Trial. To account for self-selection of participants into the

later on in the discussion section.

Randomised Controlled Trial and Data Collection

Sources: Own illustration.

data collection at t=0 had finished.

Outcomes. We collected data on two categories of outcomes: self-reported outcomes

t=0, t=1, and t=2.

Items on subjective wellbeing covered evaluative (life satisfaction), experiential (hap-

three-point seven-item Generalised-Anxiety-Disorder Questionnaire, GAD-7). PHQ-9 scores

Biomarkers included, besides cortisol, pro-inflammatory cytokines IL-1 and IL-6,

been shown to be responsive to both short-term and long-term psychosocial interventions

University of California at Irvine – using multiplex immunoassays. Cortisol was measured in

removed outliers, and log-transformed and standardised the data.

Controls. We collected survey data on socio-demographic characteristics of respond-

measured pre-treatment. Table 1b in the Supplementary Materials shows variable definitions

founding events during the study period.

Average Scores of Groups at Different Points in Time

Life Satisfaction PHQ-9 (Depression)

Treatment Group Control Group Treatment Group Control Group

GAD-7 (Anxiety) Social Trust

Treatment Group Control Group Treatment Group Control Group

with our third hypothesis.

ference-in-differences specification that compared the evolution of course outcomes between

groups over time:11

yit = 0 + 1Treatmenti*Postt + 2Treatmenti + 3Postt + 4'Xit + μs + εit with t = {0, 1}

control group anymore.

hypotheses on positive impacts on wellbeing and pro-social attitudes.

Impacts on Self-Reported Outcomes: Subjective Wellbeing, Mental Health, and Pro-Sociality

pacts were significant at the 5% level.

and anxiety. Impacts were again significant at the 5% level.