Electoral Studies 78 (2022) 102491
Available online 28 June 2022
0261-3794/© 2022 Elsevier Ltd. All rights reserved.
Too tired to vote: A multi-national comparison of election turnout with
sleep preferences and behaviors
Aleksander Ksiazkiewicz a,*
, Fatih Erol b
a
University of Illinois at Urbana-Champaign, 304C David Kinley Hall, 1407 W Gregory Drive, Urbana, IL, 61801, USA
b
Koç University, Rumelifeneri, Sarıyer Rumeli Feneri Yolu, 34450, Sarıyer, ˙Istanbul, Turkey
A R T I C L E I N F O
Keywords:
Participation
Elections
Turnout
Sleep duration
Chronotype
Multilevel context
A B S T R A C T
Receiving a healthy amount of sleep is essential to one’s quality of life. Both sleep-wake timing preferences
(chronotype) and sleep duration are implicated in health, academic achievement, and workplace performance.
This study complements the existing sleep–politics literature by examining the associations between sleep
duration, chronotype, and turnout with a representative cross-national survey dataset from nine national contexts.
Our analyses demonstrate that greater sleep duration is non-linearly related to higher turnout; those who
sleep too little or too much are less likely to vote. The results also show that morning chronotype is associated
with higher turnout, but controlling for religiosity attenuates this relationship. We argue that healthy sleep
duration and chronotype lay at the intersection of the socioeconomic and psychological resources necessary to
participate in elections.
1. Introduction
Conventional explanations for political participation focus on resources
(Brady et al., 1995), identity (Miller et al., 1981), values and
ideology (Radcliff, 1994), and social networks (Lake and Huckfeldt,
1998), with recent work adding a focus on indirect influences like implicit
cognition (Drake et al., 2015), personality (Gerber et al., 2011a,
2011b), and genetic predispositions (Klemmensen et al., 2012). We
argue that access to sleep and sleep-wake timing preferences (chronotype)
lay at the intersection of the socioeconomic and psychological
resources necessary to participate in politics.
Sleep deficits have broader social implications such as decreased
vigilance and higher accident rates (Dinges et al., 1997), lower levels of
empathy (Killgore et al., 2008), risky decision making (Ferrara et al.,
2015), less prosocial behaviors (Dickinson and McElroy, 2017), and
weakening inter-personal communication skills (Holding et al., 2019).
Eveningness-preference (late to bed, late to rise) also interacts with
morning-oriented social norms (Wittmann et al., 2006) to impair
physical and mental health, sleep quality, and workplace performance
(Fabbian et al., 2016). Prior research has found a positive relationship
between morning chronotype (early to bed, early to rise) and conservatism
in the United States (U.S.) and the United Kingdom (UK)
(Ksiazkiewicz, 2020) as well as in the context of a larger international
sample (e.g., Canada, the UK, the U.S., and also fifty-four other countries
with smaller numbers) (Ksiazkiewicz, 2021), and shorter sleep duration
and lower voter turnout in the U.S. (Holbein et al., 2019; Schafer and
Holbein, 2019; Urbatsch, 2014). Considering sleep’s centrality in our
personal and social lives and the potential importance of social institutions
in moderating those influences, we explore how sleep duration
and chronotype are related to political participation, particularly election
turnout, across national contexts.
Using nationally representative data collected in nine countries
(Finland, Greece, Ireland, Mexico, the Netherlands, New Zealand, the
Philippines, Russia, and South Korea), we integrate sleep research into
the comparative analysis of election turnout. Whereas prior work in the
U.S. suggested that inadequate sleep could lower turnout (Holbein et al.,
2019), we find that the effect of sleep duration is non-linear in seven
countries (e.g., Finland, Greece, Ireland, Mexico, the Netherlands, the
Philippines, and South Korea). In fact, excessively sleepy participants
are more likely than those with sleep deficiency to not turn out, implying
that immoderate sleep debt and abundance may differentially harm
participatory potential. We also observe that morning chronotype is
associated with higher turnout in all nine countries, but this effect is
confounded by religious observance – a civic participation covariate
influencing both sleep-wake timing preference and voting. We test the
robustness of our findings by accounting for several factors such as
* Corresponding author.
E-mail addresses: aleksks@illinois.edu (A. Ksiazkiewicz), ferol15@ku.edu.tr (F. Erol).
Contents lists available at ScienceDirect
Electoral Studies
journal homepage: www.elsevier.com/locate/electstud
https://doi.org/10.1016/j.electstud.2022.102491
Received 29 October 2021; Received in revised form 7 June 2022; Accepted 14 June 2022
Electoral Studies 78 (2022) 102491
2
day-season grouping of sleep-wake measures, religious denominations,
latitude, longitude, and mean temperature of the residential region and
the season at time of election, differences in the democratic quality of
political institutions, and time since the last election to account for
plausible turnout misreporting because of memory failures. We
conclude by laying out a sleep – political participation research agenda.
We believe our study’s theoretical perspective, empirical design, and
findings can have significant implications for understanding democratic
political participation. In principle, citizens should have an equal chance
of participating in politics, at least by voting, and the state is expected to
reduce undue burdens, for example, by introducing new tools like evoting
and postal voting for those who are unable to reach the ballot
boxes due to disability and health-related problems (Petitpas et al.,
2021). Nevertheless, alleviating the physical inaccessibility of the polling
booth does not ensure equality of opportunity in turnout. For
instance, resource inequalities are known to play an important role in
access to the ballot box (Brady et al., 1995). Despite this, sleep’s omission
from models of turnout means that nearly all past turnout research
is premised on the unstated assumption that healthy sleep in inconsequential
for turnout. Consequently, past research has ignored the heterogeneity
in access to healthy sleep in the mass public and sleep’s vital
role as a resource for political participation, even though sleep is
essential for survival and cognitive and physical performance in many
other domains of life.
2. Theory
We argue that sleep should be considered a political resource
alongside time, skills, and money (Verba et al., 1978). Unlike political
knowledge and wealth, sleep cannot be banked in advance and accumulated
over time (although the health consequences of atypical sleep
habits can accumulate over time and, in return, could dampen participatory
potential) (Burden et al., 2017). Sleep is more similar to time,
insofar as it is (or should be) replenished daily. Where it differs is that
unlike time, where everyone receives the same amount each day, access
to sleep is not equal across individuals due to personal circumstances
and structural inequalities. For instance, not every individual has synchronicity
between their individual sleep-wake cycle and the solar clock
(day-night cycle) or the social clock (societal expectations around time
use), partly owing to the variation in human physiology (e.g., age, genetics,
and sex) and partly because of the differences in the environment
(e.g., artificial lights at night, geographical and seasonal effects on
exposure to natural light at night, parenting practices, and working
hours) (Kalmbach et al., 2017). This means that those citizens not getting
enough sleep (or getting too much sleep) and/or not being able to
adjust their biological clock to the social clock will be less likely to be
energized by the mobilization of the leaders, movements, and parties to
turn out to vote, will be more likely to see the act of turnout as costly,
and will be less likely to be cognitively involved and interested in politics,
which limits their ability to to sort their individual preferences
along ideological-partisan lines and then to influence public life by
voting accordingly (cf. Smets and van Ham, 2013).
2.1. Sleep duration and voting
Previous research on how inadequate sleep may affect civic participation
(Holbein et al., 2019) lays out several direct and indirect theoretical
pathways. Firstly, those citizens lacking sleep may try to conserve
their limited energy for more personally beneficial activities than
time-costly civic participation, like turnout. Second, insufficient sleep
negatively affects citizens’ mental and physical health and thereby
might reduce prosocial behaviors such as voting because of impaired
cognitive functioning and fatigue. In addition to these theoretical expectations,
we believe that unhealthy sleep patterns may reduce electoral
participation for a variety of other direct and indirect reasons. For
instance, sleep loss, which weakens temporal memory (Harrison and
Horne, 2000), may reduce voter turnout by causing citizens to forget
election day or poll opening and closing times, vote center location,
and/or voter identification documents they must bring with them before
voting. Furthermore, even if those sleep-deprived citizens would want to
vote, insufficient sleep and the resultant vigilance decrement might
indirectly lower turnout because of hospitalizations by car accidents and
household and work-related injuries (Uehli et al., 2014). Additionally,
though it may be valid for half of the population only, sleep deprivation
may cause adult males to avoid dominance competitions such as elections
(Stanton et al., 2009) by lowering testosterone levels (O’Byrne
et al., 2021).
Accordingly, shorter sleep duration has been empirically linked with
lower voter turnout. This alignment was found using observational
cross-sectional survey analysis in the U.S. and Germany, geographical
regression discontinuity application in the U.S., and experimental
manipulation of short-term sleep in the U.S. (Holbein et al., 2019;
Schafer and Holbein, 2019), and correlational analyses on state-level
autumnal clock shifts before the U.S. general elections (Urbatsch,
2014). Considering this evidence, we pre-registered the hypothesis that
greater sleep duration may be associated with higher turnout across a
wider range of national contexts (but see section 5.3 for the additional
theoretical argument and analysis that substantiates a curvilinear relationship
between sleep duration and turnout that we did not anticipate a
priori).
H1. Greater sleep duration and higher turnout are positively related.
2.2. Chronotype and voting
Prior research has found a positive relationship between morning
chronotype and conservatism in several samples in the U.S. and the UK
(Ksiazkiewicz, 2020), as well as a positive association of conservatism
with both self-perception of being a morning type and having favorable
attitudes toward morning compared to night in a larger multi-national
convenience sample (Ksiazkiewicz, 2021). The relationship between
chronotype and political participation remains undertheorized.
Following the resource model of political participation (Verba et al.,
1995) and past sleep duration and political participation research
(Holbein et al., 2019; Schafer and Holbein, 2019; Urbatsch, 2014), it
may be reasonable to consider adequate sleep as a type of resource that
is a necessary condition to take part in politics (see section 2.1). Past
research suggests that evening types receive less sleep than morning
types because of professional pressures to wake up earlier than they
would otherwise prefer to (Giannotti et al., 2002). This is because the
biological clock of evening types, resulting in sleep-wake preference of
late to bed and late to rise, is not in sync with the social clock (e.g.,
morningness-oriented 9-to-5 business and school hours). Evening types’
sleep habits, therefore, deviate from the social clock and lead to lower
amounts of sleep on weekdays compared to morning types (Roepke and
Duffy, 2010). Combined with the observation that evening types try to
make up this sleep loss on weekends, this phenomenon is called social jet
lag (Wittmann et al., 2006). Given that sleep loss is linked to lower
turnout and that morning people are less likely to get inadequate sleep,
we might expect morningness to be connected to higher turnout.
Moreover, since morningness has a positive relationship with the Big
Five personality trait Conscientiousness, which explains individuals’
propensity for norm-compliance and rule-abidance (Lipnevich et al.,
2017), morning types might be more likely to see voting as a civic duty
to be complied with. Similarly, morningness’ positive association with
mental and physical health (Partonen, 2015) and academic achievement
and workplace performance (Tonetti et al., 2015) might make morning
types more psychologically and socio-economically well-equipped and
motivated to influence the governance of their polities by voting. Thus,
based on these previous works, we hypothesize that morningness may be
associated with higher turnout.
H2. Morningness and higher turnout are positively related.
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
3
2.3. Light pollution as the confounder between sleep and political
participation
We entertain the possibility that sleep duration and preferences and
voting are all influenced by an underlying confounding factor. Natural
and artificial lights at night affect the human circadian clock, regulating
the body’s perception of day and night (Boivin et al., 1996). As such,
light pollution in cities has been shown to disrupt the synchronicity
between the sleep-wake and day-night cycles (Wright et al., 2013). As a
result, city dwellers tend to sleep later and wake up later, with plausibly
shorter sleep duration (Pilz et al., 2018). Professional pressures,
particularly in agricultural and livestock work, may also cause people to
rise earlier in rural populations, with higher levels of sleep (Carvalho
et al., 2014). At the same time, rural residents are more likely to vote
because of the social pressures of being seen voting in small and connected
communities (Funk, 2010), possibly lower information costs of
better knowing the candidates and their policy proposals in rural areas
(Geys, 2006), the advantage of frequently using private transportation in
rural areas on election days with bad weather (Garcia-Rodriguez and
Redmond, 2020), and the relative lack of leisure time (due to nontraditional
working shifts unlike 9-to-5 weekday schedule) potentially
preventing last-minute entertainment distractions from civic affairs
(Potoski and Urbatsch, 2017). Thus, we think that urban-rural area of
residence (as a proxy for light pollution) may confound the relationship
between sleep duration and preferences and voting.
H3. The relationship between greater sleep duration and higher
turnout is attenuated by controlling for the residential area’s
urbanization.
H4. The relationship between morningness and higher turnout is
attenuated by controlling for the residential area’s urbanization.
3. Methods
We use data from two different original sources. The first is a multinational
survey study, the International Social Survey Programme (ISSP)
2007 Leisure Time and Sports module (ISSP Research Group, 2009). This
ISSP module surveys the frequency of several leisure time activities and
the leading causes for practicing and not practicing them in 34 countries.
In the module, we only use the subset of the broader international
sample with the available sleep-related variables: Finland (2007, N =
1265), Ireland (2008, N = 1975), Mexico (2008, N = 1527), the
Netherlands (2008, N = 859), New Zealand (2007, N = 948), the
Philippines (2008, N = 1092), Russia (2007, N = 1811), and South
Korea (2007, N = 1329). The second is a post-election survey study in
Greece (2020, N = 750) conducted for the Comparative Study of Electoral
Systems (CSES) from December 2019 to February 2020. Most of the
Greek CSES data (over 93%) were collected after the indirect presidential
election held by the Hellenic Parliament on January 22, 2020,
and the survey field ended before the first confirmed case of COVID-19
in Greece on February 26, 2020.1
3.1. Dependent variable
Our dependent variable is Turnout. We coded those respondents who
voted in the last elections as 1 and those who did not vote as 0. This
measure is based on direct self-reports of the participants. The elections
in consideration per country are as follows: Finland (March 18, 2007,
Parliamentary Elections), Greece (July 7, 2019, General Elections),
Ireland (May 24, 2007, General Elections), Mexico (July 2, 2006, Presidential
Elections), the Netherlands (March 7, 2006, Local Council
Elections), New Zealand (September 7, 2005, General Elections), the
Philippines (May 10, 2004, General Elections), Russia (March 14, 2004,
Presidential Elections), and South Korea (May 31, 2006, Local
Elections).
3.2. Independent variables
Our main independent variables of interest are Sleep Duration and
Chronotype (mid-point timing of sleep). We employ the variables
measuring the respondents’ times of getting up and going to bed,
ranging between 0 (00:00 o’clock) and 2359 (23:59 o’clock). While the
ISSP module inquires about sleep and wake times the day before the
interview, the Greek CSES data asks sleep and wake times when
considering free days (e.g., a normal day without work, school, or any
events to attend). The sleep measures from the ISSP module are suboptimal
because the sleep variables are from the previous day rather
than on free days, which is the conventional approach (see Research
Note 1 in the Online Appendix). However, for two reasons, we think this
measurement discrepancy might not have a bearing on our findings’
reliability. Firstly, results linking political characteristics to chronotype
are robust to different measurement approaches (see Tables S1 and S20
in Ksiazkiewicz, 2020). Second, our multivariate findings from Greece
using the conventional measure do not differ from the general trends in
countries surveyed by the ISSP.
From the available sleep measures, then, we create four variables:
Wake Hour (0–23), Wake Minute (0–59), Sleep Hour (0–23), and Sleep
Minute (0–59). Later, we develop the final constructs: Wake Timing
(Wake Hour + Wake Minute/60) and Sleep Timing (Sleep Hour + Sleep
Minute/60). We, then, calculate Sleep Duration by Wake Timing – Sleep
Timing. If Wake Timing is smaller than Sleep Timing (i.e., because Sleep
Timing is before midnight and Wake Timing is after midnight), we replace
Sleep Duration values with 24 – Sleep Timing + Wake Timing. This gives us
the number of hours slept.
Next, we calculate Chronotype by Wake Timing – Sleep Duration/2. If
the resultant Chronotype value is smaller than 0 (indicating a sleep
midpoint before midnight), we add 24 to the mid-point timing of sleep
values (to return it to 24h clock time). Later, we center the resulting
Chronotype values to 20 (e.g., Chronotype – 20 [20:00, an hour between
the latest evening chronotype and the earliest morning type in the
dataset]). Finally, we multiply the final mid-point timing of sleep
numbers by − 1 such that smaller values indicate evening chronotype
(night owl) and higher values indicate morning chronotype (morning
lark).
In the dataset, as per Ksiazkiewicz (2020), we only use the data from
survey participants whose (i) time of sleeping was after 19:00 (inclusive)
or before 06:00 (inclusive), (ii) time of waking up was after 04:00 (inclusive)
and before 15:00 (inclusive), and (iii) Sleep Duration was
in-between two and 18 h. Respondents who do not satisfy these exclusion
criteria are assumed to have atypical sleep duration and preferences
1
Even though the ISSP data was collected before our analyses, we preregistered
two of our original hypotheses about chronotype-turnout links (H2
and H4), which are not covered by the scholarship, at the Open Science
Framework (OSF) before collecting the Greek survey data (https://osf.
io/2y4ut). Even though we did not pre-register our expectation of seeing a
positive relationship between sleep duration and turnout (H1), prior theoretical
discussion and empirical evidence give us no reason to think otherwise about
this link. The confounding role of urbanization in the relationship between
sleep duration and voting (H3) is a natural extension of the pre-registered H4
because our theoretical reasoning to predict a positive relationship between
morningness and higher turnout (H4) partly benefits from the theoretical
channels on sleep duration – voting alignment (H1).
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
4
and are removed from the data.2
3.3. Control variables
We have four types of control variables that we can account for with
the available data. The first type is the geographical control, Urban-Rural
Area of Residence, the main covariate of theoretical interest. Urban-Rural
Area of Residence (H3 and H4) is on a 1–4 scale ranging from 1 (farm or
home in the country or country, village) to 2 (town or small city) to 3
(suburb, outskirt of a big city) to 4 (urban, a big city) (see Research Note
2 in the Online Appendices).
The second type consists of biological controls, Age and Sex. Age is a
ratio variable for the survey takers 18 years old or above. Sex is a binary
variable (female = 0, male = 1).
The third type involves socioeconomic controls, Education and Income.
The interval Education ranges from 0 (no formal qualification) to 1
(lowest formal qualification) to 2 (above lowest qualification) to 3
(higher secondary completed) to 4 (above higher secondary level) to 5
(university degree completed) (see Research Note 3 in the Online
Appendices). The interval Income is on a 5-point scale between 1
(lowest) to 5 (highest) (see Research Note 4 in the Online Appendices).
The fourth type comprises political controls, Level of Interest in Politics,
Left-Right Ideological Placement, and Religious Attendance. The interval
Level of Interest in Politics ranges from 1 (not at all interested) to 2
(not very interested) to 3 (fairly interested) to 4 (very interested). The
interval Left-Right Ideological Placement is on a 1–5 scale; 1 stood for “farleft,”
2 for “left, center-left,” 3 for “center, liberal,” 4 for “right, conservative,”
and 5 for “far-right” (see Research Note 5 in the Online
Appendices). The interval Religious Attendance is on a 1–6 scale; 1 stood
for “never,” 2 for “once a year or less frequently than once a year,” 3 for
“several times a year,” 4 for “once a month,” 5 for “2 or 3 times a
month,” and 6 for “several times a week, once a week, or every day” (see
Research Note 6 in the Online Appendices).
We want to include Religious Attendance, a proxy of social conservatism
potentially correlating with left-right ideology, in our models for
two reasons. Firstly, we planned to have two political controls only, Level
of Interest in Politics and Left-Right Ideological Placement. However, partly
due to greater degrees of non-response and partly due to the necessity to
use a mixture of self-placement and expert rating in ideological positioning
in the ISSP module, we wanted to test our models with a
behavioral correlate of ideology with fewer missing responses. Hence,
we have two sets of models with full controls, one with all covariates
plus Left-Right Ideological Placement and one with all covariates plus
Religious Attendance. Secondly, Religious Attendance itself can be a sole
civic participation covariate (Gerber et al., 2016). Attendance to religious
services (and subscription to conservative morality) (Zieli´nska
et al., 2021) may correlate with morningness (and also greater sleep
duration) as well as civic involvement (Polson, 2016) because of the
religious services’ regular practices adjusting the biological clocks to
social clocks and religious participation’s potential of increasing community
and duty orientation and volunteerism. All in all, religion can
theoretically be expected to be a confounding factor given the direct and
indirect theoretical pathways between morningness and religious
observance, such as the positive relationship between religiosity and
conservatism (Jost et al., 2014), morningness and conservatism (Ksiazkiewicz,
2020, 2021), morningness and the two motivational types of
values (conservation and self-transcendence) positively correlating with
religiosity (Vollmer and Randler, 2012), and morningness and the personality
trait Conscientiousness positively related to norm-compliance
and conservatism (Gerber et al., 2011a, 2011b).
Table 1 below summarizes the descriptive statistics.
3.4. Analytical strategy
To model turnout behavior, we run multilevel logistic regressions
with random intercepts and random Sleep Duration and Chronotype
slopes varying across the election types (e.g., general v. local elections)
nested into surveyed countries.3
To converge this complex non-linear
multilevel model, we use the Adaptive Gauss-Hermite Quadrature
with nAGQ = 0 (see Stegmann et al., 2018, p. 162). To reduce overcorrelation
among the control variables with fixed slopes (Paccagnella,
2006), we subtract grand means from all non-dichotomous (interval and
ratio) covariates and divide them by two standard deviations. The
dichotomous covariates are only grand mean-centered. We subtract
group means from the random slope variables (Sleep Duration and
Chronotype) and divide the resulting values by two standard deviations.
Hence, regression coefficients are comparable across models.
From our multilevel logistic regression estimations, we can obtain
three parameters: the fixed-effects (e.g., the constant effect of a variable
across individuals), the random-effects (e.g., the difference between the
fixed-effect value and the value predicted for a country), and the predicted
effects (e.g., the sum of fixed- and random-effects estimates for a
country). We look at the last parameter, the predicted effects, to test our
expectations.
4. Results
4.1. Multivariate results
The full regression models with the fixed-effects, the predicted effects,
and the random-effects are in Tables 1–3 in the Online Appendix.
The predicted effects are summarized in Fig. 1. In Model 1 with no
covariates, voting is meaningfully negatively associated with greater
duration of sleep in all nine national contexts and significantly positively
related to higher morningness in all countries except the Netherlands
(marginally) and Russia (insignificantly). While the direction of the
relationship fits our expectation for the alignment between Chronotype
and Turnout (H2), the direction of the correlation between Sleep Duration
and Turnout contradicts our prediction (H1).
Model 2 controls for Urban-Rural Area of Residence. Urbanization
does not weaken the association between Sleep Duration and Turnout in
any countries (H3) and does not attenuate the relationship between the
mid-point timing of sleep and voting (H4) in countries with significant
bivariate associations.
Comparing Model 3 and Model 4 with full controls (with either
ideology or religiosity) shows that ideological orientation confounds the
unexpected negative association between greater sleep and higher
turnout in the Netherlands and Russia (H1). Likewise, in all countries
except South Korea, religiosity confounds the expected positive link
between greater morningness and higher turnout (H2).
2
The initial sample size, the final sample size without abnormal sleep patterns
and missing sleep measures, and the total number (and percentage) of
survey takers dropped from the data due to our exclusion criteria per surveyed
country are as follows: 1,346, 1,265, and 81 (6.02%) in Finland (2007); 756
(the original sample size was 771 in which 7 respondents who were under the
age of 18 and 8 participants who had missing age information were dropped
from the outset), 750, and 3 (0.40%) in Greece (2020); 2,045, 1,975, and 70
(3.42%) in Ireland (2008); 1,558, 1,527, and 31 (1.99%) in Mexico (2008); 892,
859, and 24 (2.69%) in the Netherlands (2008); 972, 948, and 20 (2.06%) in
New Zealand (2007); 1,200, 1,092, and 102 (8.50%) in the Philippines (2008);
2,005, 1,811, and 194 (9.68%) in Russia (2007, N = 1811); and 1,431, 1,329,
and 97 (6.78%) in South Korea (2007).
3
In our model specification, next to the main country-related differences in
turnout, we want to incorporate election type specific dissimilarities as expected
by the second-order election (SOE) model (Lefevere and Van Aelst,
2014).
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
5
5. Robustness tests
5.1. Survey seasons and days
We test whether the season and day of measuring sleep-wake preferences
matter for the association between the amount of sleep, midpoint
timing of sleep and voting (Mattingly et al., 2021; Putilov et al.,
2020). Accordingly, within our multilevel models with random intercepts
and random Sleep Duration and Chronotype slopes varying across
the election types nested into the surveyed countries, we also added
random intercepts and random Sleep Duration and Chronotype slopes
varying across the days of sleep measurement, which are nested into
seasons of the survey fields (see Tables 4–6 and Research Note 7 in the
Online Appendices). We find out that additionally accounting for day
and season grouping cancels out the ideology’s confounding effect with
Sleep Duration in the Netherlands and Russia. There remains the
counter-intuitive association between greater sleep and lower turnout in
all nine countries, irrespective of covariates (H1). Further, the expected
positive relationship between greater morningness and higher Turnout
holds in all countries except Russia with and without controls (including
ideology) and in Russia when controlling urbanization or all covariates
plus ideology. Nevertheless, the season and day grouping diminish the
strong association in all countries, including the previously significant
association in South Korea, when the model has all covariates plus
religiosity (H2).
Table 1
Descriptive statistics.
Finland (N
= 1265)
Greece (N
= 750)
Ireland (N
= 1975)
Mexico (N
= 1527)
the Netherlands
(N = 859)
New Zealand
(N = 948)
the Philippines
(N = 1092)
Russia (N
= 1811)
South Korea
(N = 1329)
Overall (N
= 11,556)
Turnout: Yes (Base: No)
Mean
(SD)
[Missing
%]
0.83 (0.38)
[7.2%]
0.90 (0.30)
[0.8%]
0.72 (0.45)
[0%]
0.73 (0.45)
[4.6%]
0.91 (0.29)
[2%]
0.92 (0.27)
[9%]
0.82 (0.39)
[0.4%]
0.97 (0.17)
[8.4%]
0.68 (0.47)
[3.9%]
0.82 (0.39)
[4.1%]
Sleep Duration
Mean
(SD)
[Missing
%]
8.2 (1.5)
[0%]
7.9 (1.3)
[0%]
8.4 (1.6)
[0%]
8.0 (1.8)
[0%]
8.3 (1.6)
[0%]
8.6 (1.5)
[0%]
8.3 (1.9)
[0%]
7.9 (1.8)
[0%]
6.9 (1.7)
[0%]
8.0 (1.7)
[0%]
Chronotype
Mean
(SD)
[Missing
%]
− 7.2 (1.2)
[ 0%]
− 8.5 (1.4)
[0%]
− 7.8 (1.1)
[0%]
− 7.0 (1.2)
[0%]
− 7.6 (1.1)
[0%]
− 7.0 (1.0)
[0%]
− 6.0 (1.4)
[0%]
− 7.4 (1.2)
[0%]
− 7.5 (1.5)
[0%]
− 7.3 (1.4)
[0%]
Urban-Rural Area of Residence
Mean
(SD)
[Missing
%]
2.2 (1.0)
[0.8%]
3.0 (1.1)
[0.4%]
2.1 (1.1)
[0.9%]
2.4 (1.2)
[3.1%]
2.9 (1.1)
[0%]
2.5 (0.98)
[0.5%]
2.3 (1.3)
[0.5%]
2.7 (1.3)
[0%]
2.7 (1.0)
[0.3%]
2.5 (1.2)
[0.8%]
Age
Mean
(SD)
[Missing
%]
47 (15)
[5.2%]
45 (15)
[0%]
45 (17)
[0%]
39 (16)
[0%]
52 (17)
[0%]
51 (17)
[0%]
39 (15)
[0%]
46 (18)
[2.5%]
44 (16)
[0.1%]
45 (17)
[1%]
Education
Mean
(SD)
[Missing
%]
2.9 (1.5)
[2.1%]
3.7 (1.3)
[0.4%]
2.9 (1.3)
[0.5%]
2.4 (1.5)
[6.6%]
2.5 (1.4)
[2.7%]
3.0 (1.8)
[1.8%]
2.8 (1.4)
[0%]
3.5 (1.2)
[0%]
3.3 (1.5)
[0.2%]
3.0 (1.5)
[1.6%]
Income
Mean
(SD)
[Missing
%]
3.1 (1.4)
[18.3%]
2.6 (1.1)
[12.9%]
3.1 (1.3)
[51.5%]
3.2 (1.4)
[45.6%]
3.1 (1.4)
[7.7%]
3.2 (1.4)
[9.5%]
3.1 (1.3)
[7.1%]
3.0 (1.4)
[5.1%]
3.0 (1.4)
[4.6%]
3.1 (1.4)
[21%]
Sex: Male (Base: Female)
Mean
(SD)
[Missing
%]
0.44 (0.50)
[0%]
0.49 (0.50)
[0.8%]
0.45 (0.50)
[0%]
0.44 (0.50)
[0%]
0.48 (0.50)
[0%]
0.47 (0.50)
[0%]
0.50 (0.50)
[0%]
0.35 (0.48)
[0%]
0.45 (0.50)
[0%]
0.44 (0.50)
[0.1%]
Level of Interest in Politics
Mean
(SD)
[Missing
%]
2.2 (0.81)
[1.9%]
3.0 (0.78)
[0.7%]
2.2 (0.99)
[0.7%]
2.0 (0.98)
[3.8%]
2.7 (0.88)
[2.1%]
2.5 (0.82)
[3%]
2.3 (0.95)
[0.8%]
1.8 (0.68)
[3.6%]
2.2 (0.80)
[0.5%]
2.2 (0.92)
[2%]
Left-Right Ideological Placement
Mean
(SD)
[Missing
%]
3.0 (0.82)
[45.5%]
2.9 (1.1)
[10.7%]
3.6 (1.1)
[41.4%]
2.8 (1.2)
[73.7%]
2.7 (1.2)
[23.3%]
3.2 (0.91)
[46.2%]
3.0 (0.96)
[61.4%]
2.6 (1.4)
[81.3%]
3.0 (0.92)
[3.9%]
3.1 (1.1)
[47%]
Religious Attendance
Mean
(SD)
[Missing
%]
2.4 (1.1)
[3%]
2.9 (1.2)
[3.1%]
4.2 (1.8)
[0%]
4.6 (1.5)
[15.1%]
2.6 (1.6)
[3%]
2.5 (1.7)
[3.2%]
5.1 (1.3)
[0.3%]
2.2 (1.2)
[0%]
2.9 (2.1)
[0.5%]
3.3 (1.9)
[3.1%]
Note: Missing percentages inside the brackets reflect the missingness shares per respective sample size (N). Given our exclusion criteria on sleep patterns, Sleep Duration
and Chronotype do not have missing values.
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
6
5.2. Other individual- and country-level control variables
We test whether other individual- and country-level variables may
weaken the links between sleep and turnout behavior. We consider
whether religious denomination itself is a candidate confounder besides
religious service attendance. Prior research has linked morningness
more strongly to some religious belief systems (e.g., Roman-Catholic
denomination compared to Protestantism and atheism) (Jankowski
et al., 2014) and voting habit formation is linked to religious traditions’
time spent between worship practices and secular affairs, including
political activities (Smith and Walker, 2013). Considering the extensive
literature about how sleep preferences vary cross-nationally due to geography
(Randler et al., 2015), we also examine whether residential
region’s latitude, longitude, and mean temperature during the field of
survey and season of the turnout election are linked to the alignment
between sleep and voting. Moreover, given the different understandings
of elections (e.g., accountability of the elected officials v. resolution of
the intra-elite conflicts) as well as varying costs-benefits of voting (e.g.,
time consumption v. possibility of being oppressed) across democratic
and non-democratic regimes (Martinez i Coma, 2016), we wonder about
whether the degree of a polity’s democratic nature might structure not
only individuals’ turnout behavior but also sleep preferences and so
impair the link between the two. Lastly, we look into whether the time
between the elections (of the turnout questions) and the survey fields
may cause some respondents to underreport (or misremember) their
actual non-voting behavior, inflating the reported turnout behavior
(Kuhn et al., 2021).
Accordingly, we run our multilevel models with all main covariates
plus the categorical religious denomination, geographical-seasonal
characteristics, electoral democracy, and years since the last election
(see Tables 7–9 and Research Note 8 in the Online Appendices). The
negative association between greater sleep duration and higher turnout
is present in six out of eight countries (e.g., Finland, Ireland, Mexico,
New Zealand, the Philippines, and South Korea). As for Chronotype,
while the relation between morningness and voting is strongly positive
in three out of eight countries (e.g., Ireland, the Philippines, and South
Korea), it is strongly negative in the Netherlands, contrary to what we
expect (H2).
5.3. Non-linear conceptualization and testing
Given the existing literature on sleep duration and turnout that
informed H1, we further examined the unexpected negative alignment
between greater Sleep Duration and Turnout. Our main point of departure
is whether our (as well as the scholarship’s) linear theorizing of the
relationship between sleep duration and turnout captures the true nature
of the observed relationship. Bechler et al.’s (2021) questioning of
the common linear attitude-behavior association conceptualization led
us to think of another overlooked theoretical possibility. That is, an
excessive amount of sleep may also dampen the potential to participate
in politics.
Past research shows that oversleeping (excessive sleep duration) is
linked to impaired mental and physical health, such as memory failures
(Xu et al., 2011), anxiety disorder and depression (van Mill et al., 2014),
chronic inflammation (Grandner et al., 2013), diabetes (Shan et al.,
2015), and obesity (Chaput et al., 2008). Then, the shortest and longest
sleep duration may differentially weaken civic participation by
damaging the psychophysiological resources enabling political
involvement. Within the boundary conditions of conventional resource
theory, the negative relationship between shorter sleep duration and
voting is understandable (e.g., inadequate sleep weakening turnout
potential through lower energy). On the other hand, the health consequences
of excessive sleeping and the likely negative spillovers in
pro-social behaviors such as turnout may indicate that sleep is not
merely a socioeconomic resource with monotonic positive links to political
participation. As a result, unlike fiscal resources that increase
political participation by greater amounts, sleep duration may suppress
political participation after exceeding certain recommended ranges (e.
g., 6 to 10 h) (Hirshkowitz et al., 2015). This non-linear understanding
of the relationship between sleep and voting does not imply that the
resource theoretical framework is inappropriate for understanding
turnout in general, but rather that healthy sleep rather than sleep
Fig. 1. Predicted effects of sleep duration and chronotype on turnout.
Note: The values on the y-axis are standardized logistic regression coefficients. Those predicted effects of sleep duration and chronotype for each surveyed country are
estimated per respective model of Table 2 in the Online Appendices. There are no control variables in Model 1. Model 2 only controls urbanization. Model 3 includes
all covariates plus ideology, whereas Model 4 contains all controls plus religiosity. The thin lines are for the 99% CIs and the thick ones for the 95% CIs.
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
7
duration per se is a crucial resource for participation.
To examine the potential for a curvilinear relationship between sleep
duration and voting, we run a similar multilevel logistic regression
estimation by including second-degree polynomial term of Sleep Duration,
named Sleep Duration Squared, into the equation (see Tables 10–12
in the Online Appendices). We find that this quadratic model provides a
trend with an inverted u-shape curve where both insufficient and
excessive sleep appear to depress turnout in all countries except Russia
(see Fig. 2). Contrary to our earlier expectations, oversleeping may have
a more deleterious effect on turnout than insufficient sleep, which likely
explains the unexpected negative coefficients in the linear test of H1.
Nonetheless, the curve is least pronounced in New Zealand and Greece,
where the negative effects on turnout are primarily concentrated among
those who sleep too much rather than those who sleep too little; these
cases deserve additional attention in future research to better understand
how they differ from cases with a pronounced curve (e.g., Ireland).
To better understand the substantive significance of these findings,
we calculated effect sizes for changes in sleep duration on the probability
of voting. We compare the probability of Turnout at the peak Sleep
Duration hours (e.g., the hours below and above which turnout declines
in Fig. 2) to other sleep duration hours. For this analysis, we interpret a
roughly 1% change in turnout as a substantively meaningful effect. We
benchmark the appropriate sleep amount range between 6 and 10 h as
per Hirshkowitz et al. (2015). We take the values below and above this
recommended range as sleep deficiency and oversleeping accordingly.
Fig. 3 shows that sleep debt has a substantively negative impact on
voting in four countries (Ireland [− 4.45% to − 2.28%], Mexico
[− 1.34%], the Philippines [− 1.07%]), and Finland [− 0.96%]. We also
discover that sleeping too much also negatively affects voting in eight
countries (all except for Russia, from − 0.96% in New Zealand to
− 6.38% in Ireland). Interestingly, in four countries (Finland [− 3.22 to
− 1.09%], Ireland [− 6.38 to − 1.75%], Mexico [− 3.45 to − 1.05%], and
South Korea [− 3.64 to − 1.44%]), the negative effects of sleeping too
much begin at or near the midpoint of the recommended healthy sleep
amount (8 h). In short, the magnitude of the effect of sleep duration on
turnout is similar in magnitude to that of moving across one to two income
categories (on a five-point scale). In the same multilevel polynomial
logistic regression model, the population-level differences in the
predicted probabilities of voting at the mid-income level (with the
average predicted probability of 73.13%) compared to others are the
following (at 0.01 levels of significance): − 5.8% at the lowest-income
level, − 2.84% at the low-income level, 2.68% at the high-income
level, and 5.21% at the highest-income level.
5.4. Excluded sample
We exclude the sample with abnormal sleep patterns (e.g., sleep
duration fewer than 2 h or more than 18 h, bed time after 06:00 or before
19:00, or wake time after 15:00 or before 04:00) from the main analyses,
as applied by Ksiazkiewicz (2020). The goal of these exclusion criteria
was to eliminate respondents with extreme sleep patterns (e.g.,
extremely excessive and insufficient sleep and/or circadian rhythms
potentially symptomatic of sleep disorders such as irregular sleep-wake
syndrome and/or shift work that does not reflect their underlying sleep
preferences) or whose responses reflect measurement error rather than
their actual sleep behaviors.
Of the initial pooled sample (N = 12,205), 622 participants are
removed by the exclusion criteria and 27 respondents are also dropped
Fig. 2. Fitted values of turnout by sleep duration and sleep duration squared.
Note. The most complex model specification for this multilevel polynomial logistic regression is with random intercepts and random Sleep Duration and Chronotype
slopes varying across the election types nested into the surveyed countries and random intercepts and random Sleep Duration and Chronotype slopes varying across the
days of sleep measurement, which are nested into seasons of the survey fields. The covariates are Urban-Rural Area of Residence, Age, Education, Income, Sex, Level of
Interest in Politics, and Religious Attendance. To obtain the fitted values of turnout by sleep duration and sleep duration squared per country from Model 1 of Table 10
in the Online Appendices, we keep the remaining control and grouping variables at their (population) mean values (e.g., 0s). The dashed line is the linear smooth line
with 95% CI, and the solid line is the squared smooth line with 95% CI. Min. and Max. mean minimum and maximum values, respectively, μ denotes mean value, and
SD stands for standard deviation. The relationship between Sleep Duration and Turnout is significant curvilinearly in all countries except Russia (p > 0.10).
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
8
due to missing sleep measures (see f. n. 2). Within the excluded sample
(N = 622), 7.07% of the respondents are dropped because of sleep and
wake timing and sleep duration, 25.88% due to sleep and wake timing,
47.91% owing to sleep timing and duration, 18.81% for sleep timing
only, and 0.32% since wake timing only. Fig. 1 in the Online Appendices
shows the distribution of sleep patterns across the excluded and included
samples. Table 13 in the Online Appendices shows the percentages of
exclusion criteria used to eliminate participants across countries.
In light of our findings, another question could be raised about
whether including the previously excluded sample with abnormal sleep
patterns in the analyses would lead to major changes in the results.
Then, we run our main models by including the excluded sample (see
Table 15 in the Online Appendices). Including these participants in the
analysis leads to similar conclusions for chronotype and turnout
(generally significant effects attenuated in most contexts by religiosity)
and for the effect of controlling for urbanization (very little). For sleep
duration and turnout, we see the same negative linear relationship as
before, although more attenuated by the inclusion of controls than in
Fig. 1 (rendered non-significant in all cases except South Korea). The
non-linear relationships in Figs. 2 and 3 are particularly muddied by the
inclusion of the excluded participants (those who reported sleeping less
than 2 h, more than 18 h, or at irregular times) (see Tables 17–19 and
Fig. 2 in the Online Appendices). Regardless of whether we use our main
base hour (20:00) or much earlier timings (e.g., 16:00, 17:00, 18:00, and
19:00) for centering chronotype, the results with the excluded sample
are essentially the same (see Research Note 9 in the Online Appendices).
Where these results deviate from the analyses above (particularly for
the curvilinear effects of sleep duration), it is unclear whether these
changes are the result of measurement error introduced by the excluded
participants (thereby concealing a true curvilinear effect) or the result of
substantively different processes occurring among those participants
that should be modeled in their own right. Future research should
examine, for example, whether systematic factors such as shift work,
diagnosed sleep pathologies, or comorbidities could account for the
relationship between sleep and turnout among those with irregular sleep
patterns. Nonetheless, because most of the sample meets the exclusion
criteria and the curvilinear effect is observed across national contexts in
those cases, we believe that establishing the conditions under which this
relationship occurs should remain an area of future research.
6. Conclusion
Our study extends prior work on sleep preferences and turnout in
several ways. First, it conducts the first cross-national nine-country
analysis of sleep duration and voting, including the first such analysis for
each of those countries (as previous work focused on the U.S.). Second,
our study is the first to consider the alignment between sleep-wake
timing preference (chronotype) and voting. Third, our study is the
first to include tests of several relevant explanations, such as light
pollution, religiosity and religious denomination, the extent of electoral
democracy, time duration between the election and the interview, geography,
and non-linearity, for why sleep and political participation
may be associated.
As for our first hypothesis (H1: greater sleep duration – higher voting
Fig. 3. Differences in the predicted probabilities of turnout compared to the peak sleep duration hours.
Note. The plot is based on Model 1 of Table 10 in the Online Appendices. The percentage point differences in the predicted probability to vote are in comparison to the
peak sleep duration hours in Fig. 2 on a per-country basis (e.g., the hours below and above which turnout declines). The horizontal thin lines are for the 99% CIs and
the horizontal thick lines are for the 95% CIs. All the values are meaningful at 0.01 level of significance. CIs may not be visible because they are tight. We put the
differences in percentage values next to the respective points for illustrative purposes. The healthy sleep duration (six to 10 h), as recommended by Hirshkowitz et al.
(2015), is shaded in blue on the x-axis. The substantive effects (e.g., differences in percentage points above 1% and below − 1%) are shaded in gray on the y-axis.
Because the trend in New Zealand is heavily downward in Fig. 2, the select sleep duration value (7) is not the peak value, but rather the mid-point of the range
between 3 and 11. In Greece and New Zealand, there are no respondents with 3 h of sleep, so the respective values are missing. . (For interpretation of the references
to colour in this figure legend, the reader is referred to the Web version of this article.)
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
9
relationship), like in the U.S., getting enough sleep is associated with a
higher likelihood of turning out to vote if we compare insufficient sleep
to healthy sleep. Yet, unlike the scholarship’s orthodox linear understanding
of the association, we observe this relation only if we specify
our models curvilinearly to account for excessive sleep. From both a
theoretical and a methodological perspective, our findings suggest that
insufficient and excessive sleep might differentially diminish the likelihood
to vote in all countries except Russia (with weaker effects for
insufficient sleep in Greece and New Zealand, as well). The insignificant
opposite pattern in Russia may signal an outlier case meriting further
investigation in future research. Since Russia is the country with the
highest turnout rate in our data, we entertain three possibilities, which
are beyond the scope of our study to resolve. First, those Russian respondents
sleeping too little or too much might be imbued with fatigue
and impaired memory performance exaggerating the self-reported
turnout behavior unconsciously (Kopasz et al., 2010). Second, those
Russian participants with chronic fatigue due to sleep deficiency and
excess might be psychologically as well as socioeconomically more
susceptible to authoritarian coercion or bribery to turn out to vote (Ross,
2018). Third, the social desirability bias to report voting in Russia may
be particularly strong due to Russia’s authoritarian political system,
regardless of a respondent’s sleep patterns (cf. Kalinin, 2016). All in all,
nonetheless, understanding how over- and under-sleeping affect access
to the voting booth and what institutional reforms can address this
problem by ensuring an adequate amount of sleep or at least attenuating
this link remains a pressing comparative political behavior research
topic.
Another set of results comes from the analyses of our second hypothesis
(H2: morning chronotype – higher voting relationship). In line
with our theoretical prediction, for the first time in the scholarship, we
show that morningness is related to higher voting rates. However, the
relationship is confounded by religious observance structuring sleepwake
cycle and civic participation potential. Future multi-national
studies should prioritize higher quality measurement – for chronotype,
like the midpoint of sleep on free nights from the Munich Chronotype
Questionnaire (MCTQ) (Roenneberg et al., 2003), turnout, and some
sensitive measures such as ideology – and larger sample sizes for greater
statistical power. This would increase our confidence in these results and
facilitate explorations of how policy interventions could bolster turnout
through public education, health, and work policies that address the
socioeconomic costs of the (a)synchronicity between the sleep-wake and
the day-night cycles.
Concerning our third and fourth hypotheses (H3 and H4: light
pollution as a confounder of the sleep duration/chronotype – voting
relationship), urban-rural area of residence seems not to be a major
explanatory factor. Nevertheless, considering the association between
morningness and voting, future research should ponder how other
sources of light pollution, such as artificial indoor lighting (e.g., mobile
phones, tablet devices, and other screens)—destabilizing the human
circadian clock at night (Navara and Nelson, 2007)—may affect the link
between the sleep-wake timing preference and voting potential.
Sleep is a human universal psychobiological state that affects
everyone. However, due to individual genetic and environmental heterogeneity,
not everyone has the same chances of securing the harmony
between the sleep-wake and day-night cycles and healthy sleep duration.
Hence, the amount and timing of sleep is not always optimal for us
to function properly in society (Montaruli et al., 2021). While policy
recommendations may be implemented to encourage citizens to adopt
healthy sleep habits (e.g., sleeping six to 10 h) and preferences in harmony
with the social clock (e.g., the 9-to-5 workweek) (Centers for
Disease Control and Prevention, 2017), we do not know whether and to
what extent the recommended healthy sleep behaviors increase voter
turnout across countries. In this regard, examining the relationship between
the psychobiological features of sleep and turnout would help us
reduce the unexplained variation in voting next to genetic and environmental
effects (Ahlskog, 2021), provide a new ground for advocating
for public policies that consider wide-ranging health implications of
sleep on politics beyond extending polling hours and days (Urbatsch,
2017), such as improvements in work hour requirements, new light
policies advocating smart electronic device and home lightning systems
to minimize the negative effects of the artificial indoor lights at night on
the circadian rhythm, public healthcare expenditures on sleep disorders,
and promotion of physical activities to mitigate oversleeping at the
population-level, and so better understand the participatory foundations
of democracy attitudes in the future (Kostelka and Blais, 2018).
Given the links between sleep and voting, future work should also
delve further into how and why sleep duration and chronotype are
linked (or not linked) to other political participatory behavior. The link
between the amount of sleep and conventional participatory civic
behavior such as petition signing and donation to charities (Holbein
et al., 2019) is one of the scholarly and administratively informative
research lines in need of further investigation with cross-national
(associational, experimental, and/or longitudinal) data and with an
additional focus on chronotype. The relationship between sleep duration,
chronotype, and other conventional political participation acts (e.
g., attendance to campaign meetings and rallies, political party membership
and participation in the activities of political parties, and contact
with the elected officials about local or general issues) and also unconventional
political participation behavior (e.g., boycott, protest,
counter-protest, and online political participation), on the other hand, is
an unexplored territory looking for fresh theorization and data collection
endeavor and holding sizeable policy implications on education,
family, health, and work.
Although one strength of our research is that it delivers a new
perspective on the alignment between sleep preferences and behavior
and turnout by using untapped multinational contexts with representative
samples, the majority of our data rely on non-election studies from
the ISSP Research Group, 2009 Leisure Time and Sports module (ISSP
Research Group, 2009). It would be legitimate to ask whether surveying
key political measures such as turnout without standard election study
interview protocols and training would result in social desirability bias,
which may affect our findings. The existing scholarship on public
opinion provides evidence that there is the tendency of overreporting
turnout (Jackman and Spahn, 2019). Then, given the previous finding
that sleep deprivation is negatively related to the systematic error of
social desirability (Barber et al., 2013), one might wonder whether the
peak in turnout among respondents with adequate sleep on an inverted
u-shape curve is due to the sleep measure itself or the cognitive effort
energized by the adequate sleep duration to impress the interviewer. We
see that the aggregate turnout rates in the non-election ISSP module are
either no different or lower than the aggregate turnout rates in the
election studies of the Comparative Study of Electoral Systems (CSES)
(n.d.) Module 2 in the Philippines and Module 3 in Finland, Ireland, and
Mexico, surveyed for the same elections as in the ISSP data (see Research
Note 10 in the Online Appendices). This aggregate comparison in the
four most accessible countries suggests that the direct self-reports of
voting in the non-election ISSP module are not worryingly overstated.
Nonetheless, future research should seek to extend analyses on the sleep
and political participation links by controlling for social desirability bias
in voting self-reports (Morin-Chass´e, 2018).
Data availability
The replication data and code are available on Dataverse at
https://doi.org/10.7910/DVN/WIG6FR.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.electstud.2022.102491.
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
10
References
Ahlskog, R., 2021. Education and voter turnout revisited: evidence from a Swedish twin
sample with validated turnout data. Elect. Stud. 69, 102186 https://doi.org/
10.1016/j.electstud.2020.102186.
Barber, L.K., Barnes, C.M., Carlson, K.D., 2013. Random and systematic error effects of
insomnia on survey behavior. Organ. Res. Methods 16 (4), 616–649. https://doi.org/
10.1177/1094428113493120.
Bechler, C.J., Tormala, Z.L., Rucker, D.D., 2021. The attitude–behavior relationship
revisited. Psychol. Sci. 32 (8), 1285–1297. https://doi.org/10.1177/
0956797621995206.
Brady, H.E., Verba, S., Schlozman, K.L., 1995. Beyond SES: a resource model of political
participation. Am. Polit. Sci. Rev. 89 (2), 271–294. https://doi.org/10.2307/
2082425.
Burden, B.C., Fletcher, J.M., Herd, P., Jones, B.M., Moynihan, D.P., 2017. How different
forms of health matter to political participation. J. Polit. 79 (1), 166–178. https://
doi.org/10.1086/687536.
Carvalho, F.G., Hidalgo, M.P., Levandovski, R., 2014. Differences in circadian patterns
between rural and urban populations: an epidemiological study in countryside.
Chronobiol. Int. 31 (3), 442–449. https://doi.org/10.3109/07420528.2013.846350.
Centers for Disease Control and Prevention, 2017, March 2. How much sleep do I need?
https://www.cdc.gov/sleep/about_sleep/how_much_sleep.html.
Chaput, J.-P., Despr´es, J.-P., Bouchard, C., Tremblay, A., 2008. The association between
sleep duration and weight gain in adults: a 6-year prospective study from the Quebec
family study. Sleep 31 (4), 517–523.
Comparative Study of Electoral Systems (CSES). (n.d.). Data and documentation from the
CSES project. https://cses.org/data-download/download-data-documentation/.
Dickinson, D.L., McElroy, T., 2017. Sleep restriction and circadian effects on social
decisions. Eur. Econ. Rev. 97, 57–71. https://doi.org/10.1016/j.
euroecorev.2017.05.002.
Dinges, D.F., Pack, F., Williams, K., Gillen, K.A., Powell, J.W., Ott, G.E., Aptowicz, C.,
Pack, A.I., 1997. Cumulative sleepiness, mood disturbance, and psychomotor
vigilance performance decrements during a week of sleep restricted to 4–5 hours per
night. Sleep 20 (4), 267–277. https://doi.org/10.1093/sleep/20.4.267.
Drake, C.E., Kramer, S., Habib, R., Schuler, K., Blankenship, L., Locke, J., 2015. Honest
politics: evaluating candidate perceptions for the 2012 U. S. Election with the
implicit relational assessment procedure. J. Contextual Behav. Sci. 4 (2), 129–138.
https://doi.org/10.1016/j.jcbs.2015.04.004.
Fabbian, F., Zucchi, B., Giorgi, A.D., Tiseo, R., Boari, B., Salmi, R., Cappadona, R.,
Gianesini, G., Bassi, E., Signani, F., Raparelli, V., Basili, S., Manfredini, R., 2016.
Chronotype, gender and general health. Chronobiol. Int. 33 (7), 863–882. https://
doi.org/10.1080/07420528.2016.1176927.
Ferrara, M., Bottasso, A., Tempesta, D., Carrieri, M., Gennaro, L.D., Ponti, G., 2015.
Gender differences in sleep deprivation effects on risk and inequality aversion:
evidence from an economic experiment. PLoS One 10 (3), e0120029. https://doi.
org/10.1371/journal.pone.0120029.
Funk, P., 2010. Social incentives and voter turnout: evidence from the Swiss mail ballot
system. J. Eur. Econ. Assoc. 8 (5), 1077–1103.
Garcia-Rodriguez, A., Redmond, P., 2020. Rainfall, population density and voter turnout.
Elect. Stud. 64, 102128 https://doi.org/10.1016/j.electstud.2020.102128.
Gerber, A.S., Gruber, J., Hungerman, D.M., 2016. Does church attendance cause people
to vote? Using blue laws’ repeal to estimate the effect of religiosity on voter turnout.
Br. J. Polit. Sci. 46 (3), 481–500. https://doi.org/10.1017/S0007123414000416.
Gerber, A.S., Huber, G.A., Doherty, D., Dowling, C.M., 2011a. The big five personality
traits in the political arena. Annu. Rev. Polit. Sci. 14 (1), 265–287. https://doi.org/
10.1146/annurev-polisci-051010-111659.
Gerber, A.S., Huber, G.A., Doherty, D., Dowling, C.M., Raso, C., Ha, S.E., 2011b.
Personality traits and participation in political processes. J. Polit. 73 (3), 692–706.
https://doi.org/10.1017/S0022381611000399.
Geys, B., 2006. Explaining voter turnout: a review of aggregate-level research. Elect.
Stud. 25 (4), 637–663. https://doi.org/10.1016/j.electstud.2005.09.002.
Giannotti, F., Cortesi, F., Sebastiani, T., Ottaviano, S., 2002. Circadian preference, sleep
and daytime behaviour in adolescence. J. Sleep Res. 11 (3), 191–199. https://doi.
org/10.1046/j.1365-2869.2002.00302.x.
Grandner, M.A., Buxton, O.M., Jackson, N., Sands-Lincoln, M., Pandey, A., JeanLouis,
G., 2013. Extreme sleep durations and increased C-reactive protein: effects of
sex and ethnoracial group. Sleep 36 (5), 769–779E. https://doi.org/10.5665/
sleep.2646.
Harrison, Y., Horne, J.A., 2000. Sleep loss and temporal memory. Q. J. Exp. Psychol. Sec.
A 53 (1), 271–279. https://doi.org/10.1080/713755870.
Hirshkowitz, M., Whiton, K., Albert, S.M., Alessi, C., Bruni, O., DonCarlos, L., Hazen, N.,
Herman, J., Katz, E.S., Kheirandish-Gozal, L., Neubauer, D.N., O’Donnell, A.E.,
Ohayon, M., Peever, J., Rawding, R., Sachdeva, R.C., Setters, B., Vitiello, M.V.,
Ware, J.C., Hillard, P.J.A., 2015. National Sleep Foundation’s sleep time duration
recommendations: methodology and results summary. Sleep Health: J. National
Sleep Found. 1 (1), 40–43. https://doi.org/10.1016/j.sleh.2014.12.010.
Holbein, J.B., Schafer, J.P., Dickinson, D.L., 2019. Insufficient sleep reduces voting and
other prosocial behaviours. Nat. Human Behav. 3 (5), 492–500. https://doi.org/
10.1038/s41562-019-0543-4.
Holding, B.C., Sundelin, T., Lekander, M., Axelsson, J., 2019. Sleep deprivation and its
effects on communication during individual and collaborative tasks. Sci. Rep. 9 (1),
3131. https://doi.org/10.1038/s41598-019-39271-6.
ISSP Research Group, 2009. International Social Survey Programme: Leisure Time and
Sports—ISSP 2007 (ZA4850 Data File Version 2.0.0.). https://doi.org/10.4232/
1.10079.
Jackman, S., Spahn, B., 2019. Why does the American national election study
overestimate voter turnout? Polit. Anal. 27 (2), 193–207. https://doi.org/10.1017/
pan.2018.36.
Jankowski, K.S., Vollmer, C., Linke, M., Randler, C., 2014. Differences in sun time within
the same time zone affect sleep–wake and social rhythms, but not morningness
preference: findings from a Polish–German comparison study. Time Soc. 23 (2),
258–276. https://doi.org/10.1177/0961463X14535911.
Jost, J.T., Hawkins, C.B., Nosek, B.A., Hennes, E.P., Stern, C., Gosling, S.D., Graham, J.,
2014. Belief in a just God (and a just society): a system justification perspective on
religious ideology. J. Theor. Phil. Psychol. 34 (1), 56–81. https://doi.org/10.1037/
a0033220.
Kalinin, K., 2016. The social desirability bias in autocrat’s electoral ratings: evidence
from the 2012 Russian presidential elections. J. Elections, Public Opin. Parties 26
(2), 191–211. https://doi.org/10.1080/17457289.2016.1150284.
Kalmbach, D.A., Schneider, L.D., Cheung, J., Bertrand, S.J., Kariharan, T., Pack, A.I.,
Gehrman, P.R., 2017. Genetic basis of chronotype in humans: insights from three
landmark GWAS. Sleep 40 (2), zsw048. https://doi.org/10.1093/sleep/zsw048.
Killgore, W.D.S., Kahn-Greene, E.T., Lipizzi, E.L., Newman, R.A., Kamimori, G.H.,
Balkin, T.J., 2008. Sleep deprivation reduces perceived emotional intelligence and
constructive thinking skills. Sleep Med. 9 (5), 517–526. https://doi.org/10.1016/j.
sleep.2007.07.003.
Klemmensen, R., Hatemi, P.K., Hobolt, S.B., Petersen, I., Skytthe, A., Nørgaard, A.S.,
2012. The genetics of political participation, civic duty, and political efficacy across
cultures: Denmark and the United States. J. Theor. Polit. 24 (3), 409–427. https://
doi.org/10.1177/0951629812438984.
Kopasz, M., Loessl, B., Hornyak, M., Riemann, D., Nissen, C., Piosczyk, H.,
Voderholzer, U., 2010. Sleep and memory in healthy children and adolescents – a
critical review. Sleep Med. Rev. 14 (3), 167–177. https://doi.org/10.1016/j.
smrv.2009.10.006.
Kostelka, F., Blais, A., 2018. The chicken and egg question: satisfaction with democracy
and voter turnout. PS Political Sci. Polit. 51 (2), 370–376. https://doi.org/10.1017/
S1049096517002050.
Ksiazkiewicz, A., 2020. Conservative larks, liberal owls: the relationship between
chronotype and political ideology. J. Polit. 82 (1), 367–371. https://doi.org/
10.1086/705927.
Ksiazkiewicz, A., 2021. Political ideology and diurnal associations: a dual-process
motivated social cognition account. Polit. Life Sci. 40 (1), 56–71. https://doi.org/
10.1017/pls.2021.4.
Kuhn, P.M., Mellish, S., Vivyan, N., 2021. Psychological threat and turnout misreporting.
Elect. Stud. 70, 102276 https://doi.org/10.1016/j.electstud.2020.102276.
Lake, R.L.D., Huckfeldt, R., 1998. Social capital, social networks, and political
participation. Polit. Psychol. 19 (3), 567–584. https://doi.org/10.1111/0162-
895X.00118.
Lefevere, J., Van Aelst, P., 2014. First-order, second-order or third-rate? A comparison of
turnout in European, local and national elections in The Netherlands. Elect. Stud. 35,
159–170. https://doi.org/10.1016/j.electstud.2014.06.005.
Lipnevich, A.A., Cred`e, M., Hahn, E., Spinath, F.M., Roberts, R.D., Preckel, F., 2017. How
distinctive are morningness and eveningness from the Big Five factors of
personality? A meta-analytic investigation. J. Pers. Soc. Psychol. 112 (3), 491–509.
https://doi.org/10.1037/pspp0000099.
Martinez i Coma, F., 2016. Turnout determinants in democracies and in nondemocracies.
Elect. Stud. 41, 50–59. https://doi.org/10.1016/j.
electstud.2015.11.002.
Mattingly, S.M., Grover, T., Martinez, G.J., Aledavood, T., Robles-Granda, P., Nies, K.,
Striegel, A., Mark, G., 2021. The effects of seasons and weather on sleep patterns
measured through longitudinal multimodal sensing. Npj Digital Medicine 4 (1),
1–15. https://doi.org/10.1038/s41746-021-00435-2.
Miller, A.H., Gurin, P., Gurin, G., Malanchuk, O., 1981. Group consciousness and
political participation. Am. J. Polit. Sci. 25 (3), 494–511. https://doi.org/10.2307/
2110816.
Montaruli, A., Castelli, L., Mul`e, A., Scurati, R., Esposito, F., Galasso, L., Roveda, E.,
2021. Biological rhythm and chronotype: new perspectives in health. Biomolecules
11 (4), 487. https://doi.org/10.3390/biom11040487.
Morin-Chass´e, A., 2018. How to survey about electoral turnout? Additional evidence.
J. Exp. Polit. Sci. 5 (3), 230–233. https://doi.org/10.1017/XPS.2018.1.
Navara, K.J., Nelson, R.J., 2007. The dark side of light at night: physiological,
epidemiological, and ecological consequences. J. Pineal Res. 43 (3), 215–224.
https://doi.org/10.1111/j.1600-079X.2007.00473.x.
O’Byrne, N.A., Yuen, F., Niaz, W., Liu, P.Y., 2021. Sleep and the testis. Curr. Opin.
Endocrine Metabol. Res. 18, 83–93. https://doi.org/10.1016/j.coemr.2021.03.002.
Paccagnella, O., 2006. Centering or not centering in multilevel models? The role of the
group mean and the assessment of group effects. Eval. Rev. 30 (1), 66–85. https://
doi.org/10.1177/0193841X05275649.
Partonen, T., 2015. Chronotype and health outcomes. Curr. Sleep Med. Reports 1 (4),
205–211. https://doi.org/10.1007/s40675-015-0022-z.
Petitpas, A., Jaquet, J.M., Sciarini, P., 2021. Does e-voting matter for turnout, and to
whom? Elect. Stud. 71, 102245 https://doi.org/10.1016/j.electstud.2020.102245.
Pilz, L.K., Levandovski, R., Oliveira, M.A.B., Hidalgo, M.P., Roenneberg, T., 2018. Sleep
and light exposure across different levels of urbanisation in Brazilian communities.
Sci. Rep. 8 (1), 11389 https://doi.org/10.1038/s41598-018-29494-4.
Polson, E.C., 2016. Putting civic participation in context: examining the effects of
congregational structure and culture. Rev. Relig. Res. 58 (1), 75–100. https://doi.
org/10.1007/s13644-015-0223-4.
Potoski, M., Urbatsch, R., 2017. Entertainment and the opportunity cost of civic
participation: monday night football game quality suppresses turnout in US
elections. J. Polit. 79 (2), 424–438. https://doi.org/10.1086/688174.
A. Ksiazkiewicz and F. Erol
Electoral Studies 78 (2022) 102491
11
Putilov, A.A., Poluektov, M.G., Dorokhov, V.B., 2020. Evening chronotype, late weekend
sleep times and social jetlag as possible causes of sleep curtailment after maintaining
perennial DST: ain’t they as black as they are painted? Chronobiol. Int. 37 (1),
82–100. https://doi.org/10.1080/07420528.2019.1684937.
Radcliff, B., 1994. Turnout and the democratic vote. Am. Polit. Q. 22 (3), 259–276.
https://doi.org/10.1177/1532673X9402200301.
Randler, C., Prokop, P., Sahu, S., Haldar, P., 2015. Cross-cultural comparison of seven
morningness and sleep–wake measures from Germany, India and Slovakia. Int. J.
Psychol. 50 (4), 279–287. https://doi.org/10.1002/ijop.12098.
Roenneberg, T., Wirz-Justice, A., Merrow, M., 2003. Life between clocks: daily temporal
patterns of human chronotypes. J. Biol. Rhythm. 18 (1), 80–90. https://doi.org/
10.1177/0748730402239679.
Roepke, S.E., Duffy, J.F., 2010. Differential impact of chronotype on weekday and
weekend sleep timing and duration. Nat. Sci. Sleep 2, 213–220. https://doi.org/
10.2147/NSS.S12572.
Ross, C., 2018. Regional elections in Russia: instruments of authoritarian legitimacy or
instability? Palgrave Communications 4 (1), 1–9. https://doi.org/10.1057/s41599-
018-0137-1.
Schafer, J., Holbein, J.B., 2019. When time is of the essence: a natural experiment on
how time constraints influence elections. J. Polit. 82 (2), 418–432. https://doi.org/
10.1086/706110.
Shan, Z., Ma, H., Xie, M., Yan, P., Guo, Y., Bao, W., Rong, Y., Jackson, C.L., Hu, F.B.,
Liu, L., 2015. Sleep duration and risk of type 2 diabetes: a meta-analysis of
prospective studies. Diabetes Care 38 (3), 529–537. https://doi.org/10.2337/dc14-
2073.
Smets, K., van Ham, C., 2013. The embarrassment of riches? A meta-analysis of
individual-level research on voter turnout. Elect. Stud. 32 (2), 344–359. https://doi.
org/10.1016/j.electstud.2012.12.006.
Smith, L.E., Walker, L.D., 2013. Belonging, believing, and group behavior: religiosity and
voting in American presidential elections. Polit. Res. Q. 66 (2), 399–413. https://doi.
org/10.1177/1065912912443873.
Stanton, S.J., Beehner, J.C., Saini, E.K., Kuhn, C.M., LaBar, K.S., 2009. Dominance,
politics, and physiology: voters’ testosterone changes on the night of the 2008
United States presidential election. PLoS One 4 (10), e7543. https://doi.org/
10.1371/journal.pone.0007543.
Stegmann, G., Jacobucci, R., Harring, J.R., Grimm, K.J., 2018. Nonlinear mixed-effects
modeling programs in r. Struct. Equ. Model.: A Multidiscip. J. 25 (1), 160–165.
https://doi.org/10.1080/10705511.2017.1396187.
Tonetti, L., Natale, V., Randler, C., 2015. Association between circadian preference and
academic achievement: a systematic review and meta-analysis. Chronobiol. Int. 32
(6), 792–801. https://doi.org/10.3109/07420528.2015.1049271.
Uehli, K., Mehta, A.J., Miedinger, D., Hug, K., Schindler, C., Holsboer-Trachsler, E.,
Leuppi, J.D., Künzli, N., 2014. Sleep problems and work injuries: a systematic review
and meta-analysis. Sleep Med. Rev. 18 (1), 61–73. https://doi.org/10.1016/j.
smrv.2013.01.004.
Urbatsch, R., 2014. Time regulations as electoral policy. Am. Polit. Res. 42 (5), 841–855.
https://doi.org/10.1177/1532673X14523034.
Urbatsch, R., 2017. Youthful hours: shifting poll-opening times manipulates voter
demographics. Res. Polit. 4 (3), 2053168017720590 https://doi.org/10.1177/
2053168017720590.
van Mill, J.G., Vogelzangs, N., van Someren, E.J.W., Hoogendijk, W.J.G., Penninx, B.W.J.
H., 2014. Sleep duration, but not insomnia, predicts the 2-year course of depressive
and anxiety disorders. J. Clin. Psychiatr. 75 (2), 119–126. https://doi.org/10.4088/
JCP.12m08047.
Verba, S., Nie, N., Kim, J.O., 1978. Participation and political equality. A seven-nation.
Cambridge University Press.
Verba, S., Schlozman, K.L., Brady, H.E., 1995. Voice and equality: Civic voluntarism in
American politics. Harvard University Press.
Vollmer, C., Randler, C., 2012. Circadian preferences and personality values: morning
types prefer social values, evening types prefer individual values. Pers. Indiv. Differ.
52 (6), 738–743. https://doi.org/10.1016/j.paid.2012.01.001.
Wittmann, M., Dinich, J., Merrow, M., Roenneberg, T., 2006. Social jetlag: misalignment
of biological and social time. Chronobiol. Int. 23 (1–2), 497–509. https://doi.org/
10.1080/07420520500545979.
Wright, K.P., McHill, A.W., Birks, B., Griffin, B.R., Rusterholz, T., Chinoy, E.D., 2013.
Entrainment of the human circadian clock to the natural light-dark cycle. Curr. Biol.
23 (16), 1554–1558. https://doi.org/10.1016/j.cub.2013.06.039.
Xu, L., Jiang, C.Q., Lam, T.H., Liu, B., Jin, Y.L., Zhu, T., Zhang, W.S., Cheng, K.K.,
Thomas, G.N., 2011. Short or long sleep duration is associated with memory
impairment in older Chinese: the Guangzhou biobank cohort study. Sleep 34 (5),
575–580.
Zieli´nska, A., Stolarski, M., Jankowski, K.S., 2021. Moral foundations in chronotypes:
morningness predicts conservative morality. Chronobiol. Int. 38 (8), 1143–1150.
https://doi.org/10.1080/07420528.2021.1909611.
A. Ksiazkiewicz and F. Erol