902

British Journal of Health Psychology (2020), 25, 902–911

© 2020 The British Psychological Society

www.wileyonlinelibrary.com

Brief Report COVID-19

Daily emotional well-being during the COVID-19
pandemic
Leonhard K. Lades1* , Kate Laffan2, Michael Daly3 and
Liam Delaney2
1
Environmental Policy & Geary Institute, University College Dublin, Ireland
2
School of Economics & Geary Institute, University College Dublin, Ireland
3
Department of Psychology, Maynooth University, Co. Kildare, Ireland

The COVID-19 outbreak has become one of the largest public health crises of our time.
Governments have responded by implementing self-isolation and physical distancing
measures that have profoundly impacted daily life throughout the world. In this study, we
aimed to investigate how people experience the activities, interactions, and settings of
their lives during the pandemic. The sample (N = 604) was assessed in Ireland on the 25
March 2020, following the closure of schools and non-essential businesses. We examined
within-person variance in emotional well-being and how people spend their time. We
found that while most time was spent in the home (74%), time spent outdoors (8%) was
associated with markedly raised positive affect and reduced negative emotions.
Exercising, going for walks, gardening, pursuing hobbies, and taking care of children
were the activities associated with the greatest affective benefits. Home-schooling
children and obtaining information about COVID-19 were ranked lowest of all activities in
terms of emotional experience. These findings highlight activities that may play a
protective role in relation to well-being during the pandemic, the importance of setting
limits for exposure to COVID-19-related media coverage, and the need for greater
educational supports to facilitate home-schooling during this challenging period.

Statement of contribution
What is already known on this subject?
The existing subjective well-being literature offers some insight into potential effects of the
dramatic COVID-19-induced changes in people’s daily lives on their emotional well-being. Activities
affected by movement restrictions such as spending time in nature (MacKerron & Mourato, 2013,
Global Environmental Change, 23, 992), exercising (Reed & Ones, 2006, Psychology of Sport and
Exercise, 7, 477), walking (Hanson & Jones, 2015, British Journal of Sports Medicine, 49, 710), and
supportive interpersonal interactions (Gonza & Burger, 2017, Journal of Happiness Studies, 18, 1763;
Kahneman, Krueger, Schkade, Schwarz, & Stone, 2004, Science, 306, 1776) have been associated
with enhanced well-being. In contrast, time spent alone, engaged in social media use, and caring for
children have been associated with reduced emotional well-being (Allcott, Braghieri, Eichmeyer, &

*Correspondence should be addressed to Leonhard K. Lades, School of Architecture Planning and Environmental Policy, Planning
and Env Policy Belfield, University College Dublin, Dublin 4, Ireland (email: Leonhard.lades@ucd.ie)

DOI:10.1111/bjhp.12450
 Daily well-being during Covid-19 903

Gentzkow, 2020, American Economic Review, 110, 629; Kahneman et al., 2004, Science, 306, 1776;
White & Dolan, 2009, Psychological Science, 20, 1000). While these findings offer suggestive
evidence, their applicability to the current situation – daily life during the COVID-19 pandemic – is
unclear. Moreover, previous studies of well-being during virus outbreaks have chiefly relied on
global reports of past ‘usual’ feelings and the dynamics of daily experience have been neglected.
What does this study add?

Aspects of everyday life associated with reduced positive and reduced negative affect during the
COVID-19 pandemic include the following: Time spent outdoors; Exercising; Going for a walk;
Gardening; Taking care of children; and Engaging in in-person social interactions with friends.

Aspects of everyday life associated with reduced positive and raised negative affect during the
COVID-19 pandemic include the following: Being at work; Home-schooling children; Obtaining
information about COVID-19; Work-related social interactions; and Interactions with one’s
spouse or significant other.

Background
In the rapidly changing landscape following the emergence of COVID-19, a constant
has been the request from governments across the globe for citizens to practice
physical distancing and to isolate themselves at home. Following these guidelines,
billions of people have dramatically changed their daily routines and restricted their
movements and interactions, potentially with significant welfare effects (Brooks et al.,
2020; Lima et al., 2020). Currently, little systematic evidence exists on how people
are experiencing daily life during the pandemic. In this study, we therefore asked a
sample of the Irish population to reconstruct the activities, interactions, and
emotional experiences of the previous day. We conducted the survey on March
25th, one day after the Irish government had told all non-essential business to close,
and when citizens and workplaces had been strongly encouraged to move to home-
working. The government had previously closed all schools and universities on March
12th, and the country subsequently entered a more restrictive lockdown on March
27th. Polling during the same week indicated widespread support for the restrictions
that had been put in place (Delaney & Lunn, 2020).
Set against this backdrop, the existing subjective well-being literature offers some
insight into the effects of the dramatic changes in people’s daily lives on their emotional
well-being. Activities affected by movement restrictions such as spending time in nature
(MacKerron & Mourato, 2013), exercising (Reed & Ones, 2006), walking (Hanson &
Jones, 2015), and supportive interpersonal interactions (Gonza & Burger, 2017;
Kahneman, Krueger, Schkade, Schwarz, & Stone, 2004) have been associated with
enhanced well-being.
In contrast, time spent alone, engaged in social media use, and caring for children have
been associated with reduced emotional well-being (Allcott, Braghieri, Eichmeyer, &
Gentzkow, 2020; Kahneman et al., 2004; White & Dolan, 2009). While these findings offer
suggestive evidence, their applicability to the current situation – daily life during the
COVID-19 pandemic – is unclear.
A small set of studies have shown that self-isolation and quarantine following
previous virus outbreaks (e.g., SARS, H1N1) may produce negative psychological
effects (Brooks et al., 2020). In a Chinese sample one month into the COVID-19
outbreak, the well-being of those who were highly physically active prior to the
outbreak was particularly sensitive to the severity of the outbreak in their local area
(Zhang, Wang, Rauch, & Wei, 2020). Such studies of well-being during virus outbreaks
904 Leonhard K. Lades et al.

have chiefly relied on global reports of past ‘usual’ feelings, and the dynamics of daily
experience have been neglected.
In the current study, we therefore generated a snapshot of the experiences of people
living through the COVID-19 outbreak in Ireland using the Day Reconstruction Method
(DRM). The DRM is a diary-based tool designed to collect data on the experiences a person
has on a given day, through a systematic reconstruction conducted on the following day
(Kahneman et al., 2004). Drawing on the DRM, we estimated how affective experiences
are associated with daily activities, time spent indoors/outdoors, and social interaction
during the pandemic with a view to providing evidence to inform academic and policy
debates on how the emotional consequences of self-isolation measures may be mitigated
or exacerbated.

Method
Participants
We asked a large Irish market research company to distribute the survey link to their
access panel on Wednesday, March 25th. To be eligible for the study, participants
needed to be aged 18 or above and live in the Republic of Ireland. Eligible
participants received an email from the research company that invited them to
complete our survey before 8 pm that day. The company offered a prize draw for one
prize of €250 and five prizes of €50 additional to other draws they run for regular
survey participants. The panel contained 3,500-4,000 active members, most of them
recruited face-to-face to obtain a quota-controlled basis to be representative of the
adult population. Additionally, some participants were invited to join the panel at the
end of other surveys or using snowball recruitment. A total of 604 participants
completed our survey. The sample consisted of 191 men and 413 women, with a
mean age of 47 (SD = 12). The sample was drawn from across Irish regions
(Dublin = 31.1%, rest of Leinster = 25.8%, Muster = 23.7%, and Connaught/
Ulster = 19.4%). The majority of the sample (54.8%) had an ordinary bachelor degree
or national diploma, just under half of the sample (47.02%) were in full-time
employment, and the median household income was €50,000-€59,999. Full details of
the sample demographic characteristics can be found in Table S1.

Measures
Day reconstruction method
We utilized a short version of the DRM (Kahneman et al., 2004) where participants
completed a diary documenting what they did and how they felt during a set of up to 5
sequential ‘episodes’ from a randomly allocated section of their day. Participants were
prompted to think of episodes as scenes in a movie demarcated by transitions such as
going to a different location, or ending one activity and starting another. Participants could
describe less than 5 episodes if an earlier episode was the last episode of their day.
Participants described 4.63 episodes on average so that we recorded 2795 episodes of
March 24th in total.
Positive affect scores were calculated as the average rating of three items, that is, how
happy, how calm/relaxed, and how energetic participants reported feeling during each
episode (Cronbach’s alpha = 0.84). We selected these items to ensure positive feelings
associated with different levels of arousal were assessed (Posner, Russell, & Peterson,
 Daily well-being during Covid-19 905

2005). Negative affect was calculated using six items as the average of how sad, bored,
frustrated/annoyed, lonely, worried/anxious, and overwhelmed participants were
(Cronbach’s alpha = 0.87). Once again, we picked items to ensure negative feelings
associated with a range of arousal levels were included (e.g., bored, worried/anxious) and
to capture feelings likely to be impacted by the pandemic (e.g., overwhelmed, lonely).
The affect items assessed were similar to those included in prior DRM research
(Kahneman et al., 2004) and in mood adjective checklists such as the Positive and
Negative Affect Schedule (PANAS) (Watson & Clark 1994). Emotions were rated on a
numerical scale ranging from 1 = not at all to 7 = very much.

Statistical analysis
We first examined descriptive statistics of affect levels measured while participants were
in specific locations or engaged in specific activities. Next, we focused on how affect
ratings varied within individuals using individual fixed effect specifications that control
for fixed observable and unobservable differences across individuals using the following
regression model:

Affect it ¼ ai þ ba ait þ it

where Affect is the affect rating of individual i in episode t; a is a vector representing all
activities undertaken in episode t by individual i, and the b parameters are to be estimated;
ai is the individual fixed effect; and it is the robust error term. To ease comparison with
other studies, we standardized affect levels to have a mean of zero and standard deviation
of one. We applied this basic model to the whole sample and conducted separate fixed
effects regressions examining the associations between activities, locations, and social
interactions on the one hand and affect levels on the other hand. To manage the risk of
finding false associations in our multiple testing approach, we used the Benjamini–
Hochberg method, identifying significant associations at a false discovery rate of 0.05 (see
Supplementary Materials).

Results
Our examination of how participants allocated their time showed that they spent the
majority of their day at home (73.5%) and in the presence of people from their
household (51.8%), as anticipated. The most frequently endorsed activities were
eating, watching TV/streaming, and working or studying. Participants were alone for
27.0% of the day.

Location
Individual fixed effect regression models of the relationship between the location of the
individual and affect levels at the same time showed that being outdoors or in nature is the
location with highest positive (b = .59, SE = .05, p < .01) and lowest negative affect
(b = .25, SE = .05, p < .01). Being at work is perceived as less positive and more
negative compared to being at home as is typical (Bryson & MacKerron, 2017), as shown
in Figure 1. See Table S2 for the accompanying regression coefficients.
906 Leonhard K. Lades et al.

(a) (b)
Activities Locations
Exercising Outdoors / nature
Going for a walk Somewhere else
Gardening At other peoples home
Pursuing a hobby At home
Taking care of children At work
Socialising At a shop
Drinking alcohol
−.5 −.25 0 .25 .5
Pray/worship/meditate
Eating
Commuting to work (c) Personal interactions
Internet
Other Friends
Resting/relaxing Pets
My children
Doing housework Parents/relatives
Preparing food Nobody
Spouse/significant other
Drinking Other
Watching TV/streaming Work−related
Listening to the radio −.5 −.25 0 .25 .5
Doing nothing
Using social media
(d) Remote interactions
Schooling children
Shopping
Working/studying Pets
Other
Informing myself about Covid−19 My children
Parents/relatives
−.5 −.25 0 .25 .5 Nobody
Friends
Negative Positive Spouse/significant other
affect affect Work−related
(z−score) (z−score) −.5 −.25 0 .25 .5

Figure 1. Within-person estimates of the relationship between (a) activities, (b) locations, (c) personal
interactions, and (d) remote interactions and affect levels. Each graph presents the results of separate
fixed effects regressions. Z-scores and 95% confidence intervals are presented.

Everyday activities
Exercising (b = .46, SE = .07, p < .01), going for a walk (b = .33, SE = .06, p < .01),
gardening (b = .29, SE = .09, p < .01), and pursuing a hobby (b = .23, SE = .09, p < .05)
were ranked as the most enjoyable activities (see Table 1) and associated with the largest
increase in positive affect in our within-person analyses (see Figure 1). Taking care of
children was also linked to raised positive affect (b = .21, SE = .06, p < .01) and reduced
negative feelings (b = .19, SE = .05, p < .01). In contrast, levels of negative affect were
notably higher when participants were using social media (b = .11, SE = .04, p < .05),
home-schooling children (b = .30, SE = .07, p < .01), and informing themselves about
COVID-19 (b = .27, SE = .04, p < .01). Within-person effects of all daily activities on
affect levels are shown in Figure 1 and presented in Table S2.

Social interaction
Engaging in in-person social interactions with friends was associated with raised positive
affect ratings (b = .34, SE = .10, p < .05). In line with prior research (Bryson &
MacKerron, 2017), time spent in work-related personal or remote interactions was linked
to reduced positive and increased negative affect (see Figure 1 and Table S2). Surpris-
ingly, interactions with one’s spouse or significant other were linked to significantly lower
positive affect (b = .17, SE = .05, p < .01).

Additional analyses
Sensitivity tests indicated that the study findings did not differ markedly when fixed
effects regressions included activity dummies, location, and interaction dummies
 Daily well-being during Covid-19 907

Table 1. Mean levels of positive and negative affect in each activity and the percentage of episodes where
each activity was reported

Mean Mean % of episodes

positive negative where
affecta affectb reportedc

Activities
Exercising 5.53 2.05 5
Going for a walk 5.21 2.54 7
Gardening 5.19 2.15 3
Pursuing a hobby 4.97 2.07 3
Pray/worship/meditate 4.75 2.53 1
Socializing 4.49 2.71 6
Eating 4.39 2.67 22
Taking care of children 4.34 2.92 13
Drinking 4.29 2.82 12
Other 4.26 2.68 9
Preparing food 4.26 2.76 16
Commuting to work 4.25 3.12 3
Resting/relaxing 4.22 2.70 11
Doing housework 4.22 3.06 15
Listening to the radio 4.17 2.92 8
Drinking alcohol 4.13 3.10 2
Schooling children 4.07 3.50 3
Watching TV/ Netflix 4.07 2.92 19
Internet 4.03 3.04 1
Working/studying 3.94 2.86 17
Using social media 3.83 3.10 12
Shopping 3.80 3.03 5
Informing myself about Covid-19 3.62 3.56 11
Doing nothing 3.26 3.69 2
Location
Outdoors/nature 5.51 2.16 8
At other people’s homes 4.67 2.42 1
At home 4.14 2.79 74
At work 4.1 2.97 9
Somewhere else 4.09 2.87 4
At a shop 3.73 3.01 4

a
Positive affect is the average of calm and happy affect items.; bNegative affect is the average of
overwhelmed, sad, bored, frustrated, lonely, and worried affect items.; cParticipants reported on average
2.2 activities per episode.

simultaneously (see Figure S1) or when random effects regressions were employed
(see Figures S2 and S3). We did not find evidence of systematic differences between
males and females in the relationship between activities, interactions, locations, and
affect levels.

Discussion
We present a rich snapshot of how people are experiencing the activities and settings of
their daily lives during the COVID-19 pandemic on a day when the government had
908 Leonhard K. Lades et al.

announced strict restrictions such as the closure of all non-essential businesses. Average
positive and negative affect ratings on this day were 4.47 and 2.76, respectively. Our
analyses provide the first available estimates of how within-person variation in emotional
well-being is linked to the ways people are spending their time during the outbreak. The
findings also provide important information from a population health perspective by
highlighting key issues related to everyday activities and experiences that may require
policy consideration and inform the mental health guidelines of governments and
international bodies during the crisis.
First, in line with prior research (e.g., Hanson & Jones, 2015; MacKerron &
Mourato, 2013; Reed & Ones, 2006), we find that positive affect is greatest when
outdoors and that engaging in hobbies and physically active pursuits such as
exercising, walking, and gardening are particularly positive activities. The well-being
during outdoor activities is an important public policy consideration and needs to be
viewed in light of the trade-off between population well-being and compliance with
physical distance guidance that form part of virus mitigation efforts. However, it is
not possible to infer from the current data that elevated well-being observed in
outdoor pursuits could not be attained by substitute indoor activities. Second, our
results suggest that spending time with children following the closure of schools and
childcare facilities may benefit rather than reduce parents’ emotional well-being, as
suggested by prior research (White & Dolan 2009; Kahneman et al., 2004). In
contrast, taking on the role of educator poses significant challenges, and co-ordinated
mental health and education policies in relation to home-schooling may help to
provide necessary supports to parents at this challenging time. Third, we find that
social media use and informing oneself about COVID-19 are both associated with
elevated levels of negative affect. These results suggest that setting personal limits for
news and social media consumption, as recommended by recent research (Garfin,
Silver, & Holman, 2020) and World Health Organization guidelines (WHO, 2020),
may help to safeguard people’s emotional well-being at this difficult time. Fourth, we
observed reduced emotional well-being levels during interactions with one’s spouse
or partner, which is an atypical finding (Kahneman et al., 2004) and may reflect
difficulties in adapting to major changes in household routines and responsibilities
following the outbreak.
The current study is not without limitations. While our sample included a diverse
range of age and socioeconomic groups, it likely differs in observable and
unobservable variables from the overall population: all participants had agreed to
be part of a survey panel of a market research company, they were contacted via
email and hence were Internet savvy, and they were incentivized by various lottery
prizes. While the survey company recruited the panel on a quota-controlled basis to
be representative of the adult population, our study sample was predominantly
female indicating some selection effects. Hence, it is important to work with larger
probability-based samples to build up a full profile of everyday life during the crisis,
and the current study should not be seen as a substitute for such work (Lehdonvirta,
Oksanen, R€as€anen, & Blank, 2020). However, we examine within-person effects and
did not identify evidence that the effects identified differed systematically between
males and females. Some of our results (e.g., related to exercising and home-
schooling) rely on a small percentage (e.g., ≤ 5%) of episodes. While these findings
remain significant after controlling the false discovery rate, follow-up studies are
needed to verify these associations. Our results capture a snapshot of one point in
time, precluding an understanding of whether and how people’s routines and
 Daily well-being during Covid-19 909

emotional well-being are adapting during the crisis. For instance, as people become
more accustomed to virtual interactions, well-being benefits may follow. Multi-wave
data from a range of countries are needed to provide insight into the effects of crisis
trajectories and isolation measures on well-being on a global scale.
In conclusion, distancing people from others to limit infection is a crucial public health
measure but may also pose significant mental health risks. We aimed to untangle the
relationship between everyday activities, interactions, and emotional well-being at a time
when our sample was facing significant restrictions to their daily activities. Our findings
point to everyday activities that may mitigate (e.g., outdoor activities, gardening, exercise,
pursuing hobbies) or exacerbate (e.g., social media use, home-schooling, listening to
COVID-19 news) the welfare effects of isolation. The current study also highlights the
value in monitoring people’s daily emotional well-being during the pandemic to
understand how people are faring and to inform actions that may promote well-being
and enhance the sustainability of self-isolation measures.

Acknowledgements
We thank Orla Doyle, Margaret Samahita, Diane Pelly, and colleagues at the UCD Behavioural
Science and Policy group for very helpful comments on the project. Kate Laffan is funded by
Marie Skłodowska-Curie Individual Fellowship (project name: Mind The Gap; project number:
845342). We acknowledge financial support from the Irish Department of Health.

Conflicts of interest
All authors declare no conflict of interest.

Author contributions
Leonhard Lades (Conceptualization; Data curation; Formal analysis; Investigation;
Methodology; Validation; Visualization; Writing – original draft; Writing – review &
editing) Kate Laffan (Conceptualization; Investigation; Methodology; Writing – original
draft; Writing – review & editing) Michael Daly (Conceptualization; Investigation;
Methodology; Writing – original draft; Writing – review & editing) Liam Delaney
(Conceptualization; Funding acquisition; Investigation; Methodology; Writing – original
draft; Writing – review & editing).

Data availability statement

The data, survey materials, and analysis code will be shared on the Open Science Framework.

References
Allcott, H., Braghieri, L., Eichmeyer, S., & Gentzkow, M. (2020). The welfare effects of social media.
American Economic Review, 110, 629–676. https://doi.org/10.1257/aer.20190658
Brooks, S. K., Webster, R. K., Smith, L. E., Woodland, L., Wessely, S., Greenberg, N., & Rubin, G. J.
(2020). The psychological impact of quarantine and how to reduce it: rapid review of the
evidence. The Lancet, 395, 912–920. https://doi.org/10.1016/S0140-6736(20)30460-8
910 Leonhard K. Lades et al.

Bryson, A., & MacKerron, G. (2017). Are you happy while you work? The Economic Journal, 127,
106–125. https://doi.org/10.1111/ecoj.12269
Delaney, L., & Lunn, P. (2020). No signs of lock-down fatigue in Ireland . . . yet.. Irish Times.
Retrieved from https://www.irishtimes.com/opinion/no-signs-of-lock-down-fatigue-in-ireland-
yet-1.4218932
Garfin, D. R., Silver, R. C., & Holman, E. A. (2020). The novel coronavirus (COVID-2019) outbreak:
Amplification of public health consequences by media exposure. Health Psychology, 39, 355–
357. https://doi.org/10.1037/hea0000875
Gonza, G., & Burger, A. (2017). Subjective well-being during the 2008 economic crisis: Identification
of mediating and moderating factors. Journal of Happiness Studies, 18, 1763–1797. https://doi.
org/10.1007/s10902-016-9797-y
Hanson, S., & Jones, A. (2015). Is there evidence that walking groups have health benefits? A
systematic review and meta-analysis. British Journal of Sports Medicine, 49, 710–715. https://
doi.org/10.1136/bjsports-2014-094157
Kahneman, D., Krueger, A. B., Schkade, D. A., Schwarz, N., & Stone, A. A. (2004). A survey method
for characterizing daily life experience: The day reconstruction method. Science, 306, 1776–
1780. https://doi.org/10.1126/science.1103572
Lehdonvirta, V., Oksanen, A., R€as€anen, P., & Blank, G. (2020). Social media, web, and panel surveys:
using non-probability samples in social and policy research. Policy & Internet. https://doi.org/
10.1002/poi3.238
Lima, C. K. T., de Medeiros Carvalho, P. M., Lima, I. D. A. S., de Oliveira Nunes, J. V. A., Saraiva, J. S., de
Souza, R. I., . . . Neto, M. L. R. (2020). The emotional impact of coronavirus 2019-Ncov (new
coronavirus disease). Psychiatry Research, 287, 112915. https://doi.org/10.1016/j.psychres.
2020.112915
MacKerron, G., & Mourato, S. (2013). Happiness is greater in natural environments. Global
Environmental Change, 23, 992–1000. https://doi.org/10.1016/j.gloenvcha.2013.03.010
Posner, J., Russell, J. A., & Peterson, B. S. (2005). The circumplex model of affect: An integrative
approach to affective neuroscience, cognitive development, and psychopathology. Develop-
ment and psychopathology, 17(3), 715–734. https://doi.org/10.1017/S095457940505
0340
Reed, J., & Ones, D. S. (2006). The effect of acute aerobic exercise on positive activated affect: A
meta-analysis. Psychology of Sport and Exercise, 7, 477–514. https://doi.org/10.1016/j.psyc
hsport.2005.11.003
Van den Berg, A. E., Maas, J., Verheij, R. A., & Groenewegen, P. P. (2010). Green space as a buffer
between stressful life events and health. Social Science & Medicine, 70, 1203–1210. https://doi.
org/10.1016/j.socscimed.2010.01.002
Watson, D., & Clark, L. A. (1994). The PANAS-X: Manual for the positive and negative affect
schedule-expanded form. Ames, IA: The University of Iowa.
White, M. P., & Dolan, P. (2009). Accounting for the richness of daily activities. Psychological
Science, 20, 1000–1008. https://doi.org/10.1111/j.1467-9280.2009.02392.x
WHO (2020). Mental health and psychosocial considerations during the COVID-19 outbreak. WHO/
2019-nCoV/MentalHealth/2020.1
Zhang, S. X., Wang, Y., Rauch, A., & Wei, F. (2020). Unprecedented disruptions of lives and work.
Health, distress, and life satisfaction of people one-month into COVID-19 outbreak in China.
Psychiatry Research, 288, 1–6. https://doi.org/10.1016/j.psychres.2020.112958

Received 8 April 2020; revised version received 12 May 2020

 Daily well-being during Covid-19 911

Supporting Information
The following supporting information may be found in the online edition of the article:
Table S1. Descriptive statistics for the study sample (N = 604).
Table S2. Separate within-person estimates of the relationship between activities,
locations, personal interactions and remote interactions and affect levels.
Figure S1. Mutually adjusted individual fixed effects regressions in which we added
activities, locations, in-person and remote social interactions in the same model.
Figure S2. Random effects specifications without additional controls.
Figure S3. Random effects specifications controlling for age and gender.