The Differential Effects of Stress on Voter Turnout
Hans J. G. Hassell
Cornell College
Jaime E. Settle
College of William & Mary
While everyone deals with stressful situations on a daily basis, individuals have different behavioral reactions
to that stress. We argue that life stress also affects individuals’ political behavior, but this effect is contingent
on their past political involvement. While individuals familiar with and engaged in the political process are
unaffected when confronted with stress in life, individuals who are not routinely involved in the electoral
process are more likely to disengage from politics. To test the differential effects of stress on the likelihood of
political involvement, we fielded two experiments, one preceding the U.S. presidential election of 2012 and the
second preceding the 2013 municipal election in a small Midwestern American town. We find that when
triggered to consider life stressors unrelated to politics, individuals without a history of past participation in
politics are less likely to vote while individuals who are habitual voters are unaffected.
KEY WORDS: stress, voter turnout, political participation
Every day individuals must deal with personal problems and stressful situations in their lives.
The norms of civil society require that citizens participate in the political process in spite of these
daily problems, as the function of an effective democracy depends on the participation of its citizens.
Yet most people do not fully participate in all aspects of the process, and many do not participate at
all. While previous studies of participation have focused primarily on the resources individuals have
available such as time, money, and civic skills, recent evidence finds that an individual’s well-being
and overall health also significantly influence political engagement (Mattila, S€oderlund, Wass, &
Lauri, 2013; Ojeda, in press; Pacheco & Fletcher, 2015). Indeed, stressors from a wide range of life
experiences and circumstances have a major influence on an individual’s participation in the political
process (Hobbs, Christakis, & Fowler, 2013; Pacheco & Plutzer, 2007, 2008; Sandell & Plutzer,
2005; Verba, Schlozman, & Brady, 1995).
Scholars have acted on the assumption, however, that stressful situations—and the costs they
impose—uniformly and negatively affect the likelihood that individuals participate in the political
process. This customary explanation suggests that as the life of the average American becomes more
stressful, he or she is less likely to vote, as those problems and concerns occupy the time otherwise
required to become informed about the political process and to participate. Under the assumption of
universal stress effects, life stressors either reduce the availability of resources necessary to participate,
or they increase the opportunity costs associated with voting. Differences in observed behavior
533
0162-895X VC 2016 International Society of Political Psychology
Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, 9600 Garsington Road, Oxford, OX4 2DQ,
and PO Box 378 Carlton South, 3053 Victoria, Australia
Political Psychology, Vol. 38, No. 3, 2017
doi: 10.1111/pops.12344
bs_bs_banner
thus stem from differences in the educational and economic resources that individuals have at their
disposal to pay those costs (Rosenstone & Hansen, 1993; Verba et al., 1995; Wolfinger & Rosenstone,
1980).
This assumption ignores the possibilities of the conditional effects of stress on political participation.
While the bulk of the literature has shown that stress-inducing life events lower the political participation
of the general population (Hobbs et al., 2013; Pacheco & Plutzer, 2008; Sandell & Plutzer,
2005), some studies involving certain subpopulations have shown the opposite, namely that stress in
life has no effect or even increases political participation (Pacheco & Plutzer, 2007; Peterson, 1987;
Peterson & Gabbidon, 2007; Peterson & Somit, 1994). Likewise, just as there is some limited evidence
that the adaptive family and community structures of racial minorities provide a buffer to some
of the political effects of life stressors (Pacheco & Plutzer, 2007), research in psychology has shown
that there are important differences in how people handle and cope with stress (Bandura, 1977; Lazarus
& Folkman, 1984). An individual’s choice of how to respond to the presence of stress depends on
that individual’s appraisal of the situation. Past assumptions about the uniform and negative effects of
stress do not take into consideration the means by which people deal with stress in their lives (Bandura,
1977; Lazarus & Folkman, 1984), their past political behaviors (Plutzer, 2002), or their ability
to connect personal life stress to political or governmental solutions (Brody & Sniderman, 1977).
We argue that those differences moderate how stress affects an individual’s political behavior.
For those people who are regularly involved in and familiar with the political process, life stressors do
not deter participation. For these individuals, participation does not require the time and consideration
that it does for habitual nonvoters (Blais, 2000; Plutzer, 2002). Conversely, additional stress in the
lives of nonhabitual participants decreases the likelihood of participation. For those individuals, the
act of voting merely creates additional demands and stresses to an already demanding life. Stress in
life for nonhabitual voters increases the opportunity costs associated with voting and causes them to
withdraw from politics. Unlike voters familiar with the political process, personal troubles cannot be
directly connected to the political process and, as such, increase the costs the individual must incur in
order to participate.
In short, we argue that the effects of stress on political participation are conditional on past political
behavior. To test this hypothesis, we fielded two experiments. The first of these was an online survey
experiment that measured vote intentions. While controlled laboratory experiments better allow
for the isolation of the variables of interest and ensure that respondents receive the full effects of the
treatment, the environmental circumstances are significantly different from those experienced in the
real world. Chief among those is the difference between an individual’s stated intentions and his or
her actual behavior (Abramson & Claggett, 1984; Berinsky, 2004; Katosh & Traugott, 1981). For that
reason, we also ran a similar second field experiment in a small town in the Midwestern United States
that measured actual turnout behavior. While the field experiment limits our ability to ensure the
receipt and processing of the treatment, it better allows for us to see the effects of the treatment on
actual behavior rather than stated intentions. Using a combination of field and lab experiments ensures
that both the internal and external validity of our findings are robust.
In both experiments, one group of respondents was prompted to contemplate stressful things in
their lives. The results from the first experiment suggest that compared to a control group that received
no stress stimuli, nonhabitual voters in the treatment group are less likely to indicate they would vote
and are also less likely to actually turn out to vote in the second experiment. On the contrary, habitual
voters who were asked to contemplate stressful things in their lives were no less likely to indicate that
they intended to vote in the first experiment or actually vote in the second experiment. As such, we
argue that the assumption that the costs of stress are uniformly negative is wrong. We find that habitual
voters and nonhabitual voters respond differently politically to stress in their daily lives, even after
controlling for the resources commonly used to explain participation (Verba et al., 1995).
534 Hassell and Settle
The Cost of Voting
The traditional rational choice voting calculus characterizes the decision to vote as a simple equation
where the potential voter weighs the costs and benefits of voting (Downs, 1957; Duffy & Tavits,
2008). Many voters choose not to participate because of the overwhelming costs associated with casting
a ballot. Costs to voters can take a myriad of forms, either institutional (Highton, 2004; Powell,
1986) or individual (Brady, Verba, & Scholzman, 1995), but all decrease the likelihood of participation
for individuals unable or unwilling to pay those costs.
Life Stressors and the Vote
A recently burgeoning literature on health effects has shown that one’s overall well-being (Mattila
et al., 2013; Pacheco & Fletcher, 2015), physical capacity (Schur, Shields, Kruse, & Schriner,
2002), and mental health (Ojeda, in press) all influence political participation rates of Americans.
There is a well-documented literature on the long litany of trying events in life that affect political
engagement. Among older rural Americans, stress and hardship in life reduces political interest, political
efficacy, and participation (Peterson, 1987). Likewise, stressful life events—such as widowhood
(Hobbs et al., 2013), divorce (Sandell & Plutzer, 2005), and economic hardships (Pacheco & Plutzer,
2007, 2008)—also have a dramatic negative effect on the likelihood of participation.
However, many of these studies find a wide variation in the size of the effects of these stressful
occurrences among different subgroups, showing a large dampening effect on turnout in some subgroups
and little to no effect in others (Pacheco & Plutzer, 2007). A few studies have even found that,
among some subgroups, poor health and traumatic life events lead to greater political participation
(Peterson & Gabbidon, 2007; Peterson & Somit, 1994). Political scientists have long contended that
people’s perceptions of their problems impact political behavior depending on the degree to which
those problems can be connected to the political realm, concepts operationalized in Brody and Sniderman
(1977) as the locus of concern and in Sniderman and Brody (1977) as the focus of expectations.
Analyzing data from the 1972 American National Election Study (ANES), the authors find that those
who participate in the political process on a regular basis are better able to connect the stress in their
lives to a political solution (Brody & Sniderman, 1977). Furthermore, they find evidence of an inhibitory
effect, where respondents who are concerned about personal problems (in their language, personal
economic problems or self-located problems) are less likely to vote than we would expect based
on their demographic characteristics alone.
These findings suggest that the differential effect sizes found in previous studies for the relationship
between stressful life occurrences and political participation may be due to the role of past experience
in the political realm fostering the ability to link personal problems to political solutions. A
person’s ability to view stressful events as fixable through the political process may largely depend
upon his or her familiarity with the process and how that process affects his or her life. Thus, we turn
to the literature on habitual voting to understand why personal stressors may differentially affect people
based on their prior voting history.
Habitual and Nonhabitual Voters
A large body of evidence now supports the importance of the habit-formation aspects of voting.
Voters who are mobilized to vote for the first time in one election are more likely to vote in subsequent
elections than are similar voters who were not mobilized (Coppock & Green, 2015; Gerber,
Green, & Shachar, 2003; Green & Shachar, 2000). People who have participated in the political process
in the past are more likely to participate again in the future, in part because of their characteristics
(Grant & Rudolph, 2002), but also because future electoral campaigns are more likely to target them
Differential Effects of Stress on Voter Turnout 535
in their efforts (Hassell & Monson, 2014). One’s status as a voter or nonvoter also appears to be subject
to some inertia (Milbrath, 1965), which limits the ability of citizens to move seamlessly between
groups. Individuals who are voters are likely to remain voters unless they encounter prohibitory barriers
to voting. Individuals who are nonvoters, on the other hand, must pay higher costs to overcome
the inertia of not voting, a cost that habitual voters do not normally face (Plutzer, 2002).
The concept of habit formation has come to form an important part of our understanding of why
people vote, but the role of habit formation is nuanced and complex (Coppock & Green, 2015) as it
likely intervenes and interacts with other important predictors of turnout. For example, many of the
individual-level factors that contribute to voter turnout—ranging from demographic factors to attitudinal
factors—appear to contribute to habit formation, suggesting that habit formation may play a large
mediating role in these relationships. Simultaneously, habit formation may moderate citizens’ responsiveness
to contextual-level factors thought to be associated with increased voter turnout. The citizens
most responsive to electoral competitiveness are those with little past voting experience (Franklin,
2004), and first-time voters appear to more sensitive to the effect of living with other voters (Fieldhouse
& Cutts, 2012), though neighborhood electoral context does not seem to differentially effect
first-time voters (Fieldhouse & Cutts, 2012).
The Differential Effects of Life Stress
In our study, we test the possibility of habit formation to moderate the effect of a psychological
factor—activated life stress—on voter turnout.1
How and why should habit formation moderate the
effects of activated stress on turnout? The answer to that question depends on the cognitive and emotional
consequences of activating stress, an entire field of inquiry within psychology and a topic worthy
of future investigation. However, we propose that reminding people of the stress in their daily
lives has two main consequences. The first is to remind people of the opportunity costs of voting;
every minute spent engaging in politics is time not spent addressing other financial or personal problems.
The second consequence is to activate citizens’ linkage of their personal problems to government
solutions.
These two consequences differentially affect habitual and nonhabitual voters. The act of voting
creates additional demands and stresses to an already demanding life, but this is differentially costly
based on one’s past experience with politics. Because many of the cognitive costs of voting are typically
reduced for habitual voters, activation of additional costs via a reminder of life stress is insufficient
to change the fundamental cost-benefit analyses in which a habitual voter engages. However,
stress in life for nonhabitual voters increases the opportunity costs associated with voting, and troubles
and stressors in their lives cause a withdrawal from politics. Unlike voters familiar with the political
process, personal troubles cannot be directly connected to the political process and therefore increase
the costs the individual must incur in order to participate.
Our study pulls together these two findings from two distinct literatures to argue that the experience
of being a habitual voter can moderate a factor known to explain voter turnout: stressful life
experiences. Previous studies about habitual voting have not focused on its ability to moderate the
psychological factors known to be associated with voter turnout. Nor have previous studies about the
role of life stress in political participation explored factors that explain why people are differentially
responsive to life stress.
1
We remain agnostic about whether habit formation also mediates the relationship between certain psychological predictors
and turnout: a situation in which the way that people process the stress in their lives contributes to their underlying
propensity to become a habitual voter. There is no clear prediction here in the literature. The pattern of findings
from the closest stable personality trait regularly measured in the literature—emotional stability—is mixed with
regards to various forms of political participation (Mondak, 2008; Mondak, Hibbing, Canache, Seligson, & Anderson,
2010).
536 Hassell and Settle
Furthermore, these studies examine stressors in isolation and have relied on observational data
that make it difficult to observe and isolate the process through which the addition of stress to an individual’s
life matters in his or her decision to participate in politics.2
To varying degrees, these studies
suggest a causal relationship between life stress and participation decisions, but they have not isolated
the role of stress in the turnout decision. Rather they have focused on life conditions assumed to be
stressful, without assessing the degree to which subjects themselves feel stressed by their circumstances.
By using experimental methods, we are better able to causally link stress in life to individual
political action.
Hypotheses
For regular voters, voting is a straightforward task, is more likely to be connected to economic
related stress in life, and is a habitual process; therefore, we expect that additional stress in the life of
a habitual voter will have no effect on her likelihood of voting.
H1: Reminding habitual voters about stressful things in their lives will have no effect
on their likelihood of voting.
Nonhabitual voters, as a result of their lack of familiarity with the voting process, are more likely
than habitual voters to view the process of voting as a particular burden and are also less likely to
understand the relationship between stressors in their individual life and the political process. As such,
we expect nonhabitual voters’ voting behavior will decrease when exposed to stress. To these individuals,
additional stress raises the opportunity costs associated with voting, lowering the likelihood they
will vote.
H2: Nonhabitual voters will be less likely to vote when reminded about stressful things
in their lives.
In summary, in contrast to nonhabitual voters, citizens who are familiar with the political process
are more likely to connect stress in their daily life to the political process and less likely to view the
voting process as an onerous task. Consequently, we should find that making life stress more salient
should have a negligible or positive effect on the likelihood of voting for habitual voters and a negative
effect on the likelihood of voting for nonhabitual voters.
Controlled Experiment
To test our hypothesis about the differential effect of stress on political participation, we ran two
experiments. The first of these experiments was an online survey experiment in which we recruited
1,278 subjects who lived in the United States through Amazon’s Mechanical Turk the week before
the 2012 presidential election. Amazon’s Mechanical Turk is a website where requesters publish tasks
(HITs or Human Intelligence Tasks) and provide payment to those who choose to participate. Those
who request a task can limit the availability of the task to respondents who meet certain qualifications,
such as age or location. For our experiment, we required that respondents be citizens at least 18 years
old living in the United States. Recruitment through Mechanical Turk is similar to other web-based
2
The majority of these studies are also cross-sectional designs, with the exception of the work of Pacheco and Plutzer
(2008) who use panel data for events occurring in adolescence. We know of no prior experimental work on the effects
of life stress and political participation.
Differential Effects of Stress on Voter Turnout 537
approaches such as YouGov that maintain panels of participants and invite them to participate in studies
in exchange for a payment or other incentive. Studies that have used data from Mechanical Turk
have been published in the top journals in political science (e.g., Grimmer, Messing, & Westwood,
2012; Hassell & Visalvanich, 2015). Additionally, studies run from samples collected from Mechanical
Turk replicate important experimental findings in psychology (Buhrmester, Kwang, & Gosling,
2011), and previous research has shown that scholars can collect more representative samples from
Mechanical Turk than they can from undergraduate samples or samples populated from those who
respond to web advertisements (Berinsky, Huber, & Lenz, 2012).
Other issues with convenience samples, including online convenience samples, is the concern
that treatment effects, or more specifically the lack thereof, may be driven by homogeneity of unmeasured
characteristics of the subjects in the sample. It is possible that characteristics of individuals who
take surveys online could confound the heterogeneous treatment effects; however, previous scholarship
has only found convergence of these effects rather than a divergence and only in experimental
treatments that require the subject to trust information provided by the experimenter (Krupnikov &
Levine, 2014). Other similar experiments have found no confounding of heterogeneous treatment
effects in similar experiments, including experiments using longer vignettes (Weinberg, Freese, &
McElhattan, 2014) and elaborate buy-ins on the part of respondents (Berinsky et al., 2012).
We recognize the inherent limitations of using Mechanical Turk. However, although a
sample may not be entirely representative of the general population, its usefulness is dependent
on the variation of relevant moderating characteristics (Druckman & Kam, 2011). In our
case that variable is past voter history.3
The sample from Mechanical Turk had a self-reported
turnout rate of almost 59% in the 2008 election and 41% in the 2010 midterm elections compared
to an actual voter eligible turnout rate of just under 62% in the 2008 national election
and 42% in the 2010 midterm elections.4
While self-reported turnout is generally higher than
actual turnout, the approximate similarity of these numbers suggests our sample mirrors the
national adult population on voter turnout. Table 1 compares the turnout rate of individuals in
our sample to the turnout rate in the general population in 2008 and 2010 and the demographics
and political efficacy of our sample to the sample used in the 2012 American
National Election Survey. While our survey participants are younger, slightly poorer, and
more educated than a more representative national sample, the demographic sample in our
study is more representative than convenience samples commonly used in political science
Table 1. Survey Sample Comparison
Survey Sample Actual U.S. Population
Turnout 2008 58.66% 61.60%
Turnout 2010 41.30% 41.70%
Survey Sample ANES 2012
White 79.39% 70.91%
Less Than a College Degree 54.59% 70.59%
With Income <$40K 49.95% 40.67%
Under 35 69.33% 28.95%
Average External Efficacy 20.144 20.286
3
Previous research has also found that there is no support for speculations that busy people are less likely to respond to surveys.
Busy people are just as likely to respond to surveys as those with more leisure time (Abraham, Maitland, & Bianci,
2006). While we are not able to measure how busy people were in the Mechanical Turk survey, given the consistency of
the covariates outlined in Table 1, we suspect that there is also reasonable variation in the fullness of individuals’ lives.
4
Voting Eligible Turnout was 59% and 41%; Voting Age Turnout was slightly lower. Data on turnout is from Michael
McDonald’s United States Election Project (http://www.electproject.org).
538 Hassell and Settle
and psychology. More importantly, our sample provides a conservative test of our hypothesis
that there are differential effects of stress on habitual and nonhabitual voters. Studies indicate
that habit strength is reinforced over time and as the activity is performed repeatedly (Milbrath,
1965). Thus, if we find a differential effect between habitual voters and habitual nonvoters
at a younger age when levels of inertia are lower, then it is more likely that these
differential effects would be stronger in an older population where levels of inertia keeping
individuals in their voting habits are higher.
After gathering basic information about the respondents, we implemented a single factor experiment
with three levels. Respondents were randomly assigned into three different groups with two
treatment groups and a control group. Hotelling balance tests showed that our randomization was successful
and revealed no significant differences between the groups on socioeconomic and partisanship
variables. Respondents were asked to complete a task designed to elicit a specific psychological
response, and we varied the task between subjects. There was no within-subjects manipulation. We
assigned the two treatment groups to engage in one of two different directed-thinking tasks designed
to prompt respondents to conceptualize themselves as stressed or content (Kunda, Fong, Sanitioso, &
Reber, 1993; Levine, 2015; McGuire & McGuire, 1996). Kunda et al. (1993) find that “directional
questions influence self-conceptualizations” and that these directed questions cause individuals “to
believe that they have higher levels of that attribute” (p. 82). In the first treatment, we asked respondents
to list up to three things in their life that increase their levels of stress.5
To the extent we were
able, we could find no meaningful differences between habitual and nonhabitual voters in the content
Figure 1. Indicated likelihood of voting for all experimental primes in the Mechanical Turk study. Lines indicate 95%
confidence intervals.
5
The exact wording of the question was “People often find that there are many things in their day-to-day life related to
family and employment that bring stress to their lives, such as choices their family members make, their relationship
with their spouse or partner, a difficult boss, a long commute, or an unsecure job, for example. Being as specific as
possible, please list up to three things relating to your family life, job, or occupation that add stress to your life.” Individuals
were then shown their responses and asked to confirm them.
Differential Effects of Stress on Voter Turnout 539
listed. (See the appendix in the online supporting information for a fuller explanation of this analysis.)
After listing these items, we showed them the items they had listed and asked respondents to review
them to make sure the items were correct. We designed this treatment to make stress more salient by
priming respondents to think about those things that are stressful to them.6
Previous work has found
that subjects who are asked directional questions such as these are more prone to conceptualize themselves
in that state of being because they are more likely to access memories confirming these conceptualizations
(Kunda et al., 1993).7
By reminding people of the stress in their lives we make them more
Table 2. Effect of Stress-Inducing Prime on Indicated Likelihood of Voting
(1) Likelihood of
Voting
(2) Likelihood of
Voting
(3) Likelihood of
Voting
(4) Likelihood of
Voting
Experimental Treatments
Life-Stress Treatment 3.055 3.482 3.055 3.292
(3.547) (3.467) (3.575) (3.459)
Pleasant-Thoughts
Treatment
1.342 21.508
(3.713) (3.581)
Respondent Characteristics
Nonhabitual Voter 225.918*** 220.759*** 225.917*** 220.846***
(3.205) (3.270) (3.230) (3.215)
Efficacy 6.064*** 6.681***
(1.369) (1.139)
Partisan 6.996*** 8.941***
(2.245) (1.848)
College Degree 12.129*** 11.022***
(2.942) (2.370)
Income $40K to $80K 0.150 1.544
(2.365) (1.955)
Income $80K1 8.094** 7.806***
(3.155) (2.579)
White 1.525 3.660
(2.814) (2.264)
Age 0.548 0.391
(0.575) (0.467)
Age2
20.007 20.005
(0.007) (0.006)
Interactions
Nonhabitual Voter 3
Stress Treatment
27.904* 29.262** 27.904* 29.226**
(4.496) (4.389) (4.531) (4.381)
Nonhabitual Voter 3
Pleasant Thoughts
23.578 24.186
(4.628) (4.465)
Constant 94.686*** 65.677*** 94.686*** 66.736***
(2.561) (11.618) (2.582) (9.439)
Observations 863 863 1,278 1,278
R2
0.174 0.232 0.168 0.235
RMSE 31.99 30.99 32.24 31.01
Note. OLS coefficients with standard errors in parentheses.
***p < 0.01, **p < 0.05, *p < 0.1.
6
This differs from previous work in political science that has made assumptions about the kinds of things—interpersonal
relationship change (divorce, spousal death) or financial strain (poverty rates)—that might be stressful for individuals.
Our treatment is thus the process of thinking about specific stressful things related to one’s life, which has a
direct relationship with one’s self-conceptualization (Kunda et al., 1993).
540 Hassell and Settle
likely to recover a biased set of memories that causes them to believe they are in a certain mental state
and thus affects their decision-making as if they were in that state (Levine, 2015). While we are
unable to test whether these prompts cause a physiological state of stress, they affect individuals’ perception
of stress in their lives.
However, because we were concerned that the process of thinking and considering up to three
items might increase respondents’ focus and alter evaluations of the likelihood of voting in the
upcoming election above and beyond the effect of conceptualizing stress in their lives, we prompted
the second group to think and list up to three of the “things that make your day a little brighter and
more pleasant.” As with the group that received the stress treatment, we then showed individuals the
list of items that they had listed as brightening their day and asked them to review them to make sure
that they were correct. After receiving the treatment, individuals were asked how likely they were to
vote in the upcoming election on a scale of 0–100.8
The final group acted as a control and received no
self-directed thought treatment; respondents in that group were merely asked to indicate how likely
they were to vote in the upcoming election.
Different Effects for Habitual and Nonhabitual Voters
We begin by comparing the indicated likelihood of voting of all three groups in Figure 1. The
results indicate that the mean likelihood of predicted turnout in each of the treatment groups is not significantly
different from that in the control. On the whole, priming individuals to consider stressful situations
does not significantly change their indicated likelihood to vote.
As hypothesized, however, differentiating between those who vote habitually and those who do
not vote on a consistent basis demonstrates a differential effect of stress on predicted turnout. Simply,
stress affects these groups differently. We classified habitual voters as individuals who voted in both
the 2008 presidential election and the 2010 midterm election.9
Voters who participated in only one
election or in neither election were classified as nonhabitual voters.10
To model the effect of making
stress salient on the propensity of individuals to turn out to vote, we constructed a model that controls
for a variety of other factors that have been shown to be influential in an individual’s turnout decision.
Because habit formation is not randomly assigned, it is important to also control for a variety of socioeconomic
factors that have been shown to cause habit formation (Franklin, 2004; Plutzer, 2002). This
allows us to isolate the habit of voting from the other factors that cause the formation of habit. Table 2
contains a series of models predicting an individual’s indicated voting likelihood. The first two
7
Because we prompted people to think about stress (and not anxiety) they should conceptualize themselves as stressed
(Kunda et al., 1993). However, we recognize that in the general public, the distinction between stress and anxiety is
not as well defined as it is in scholarship, and it could be that these individuals thought of themselves as anxious
rather than stressed.
8
We used a scale of 0–100 to increase the variance in our measurement to facilitate identification of effects with a
small sample and because we were concerned with ceiling effects among habitual voters using an ordinal measure of
anticipated likelihood of voting. We recognize that this makes it difficult to identify substantive effects, especially
among habitual voters, and thus in the later field experiment we use a verified measure of turnout.
9
Previous research has not been consistent in identifying how to classify habitual voters. Franklin (2004) identifies voters
as forming a voting habit after voting in three consecutive elections. We, however, are not interested in the formation
of voting habits, but rather whether the influence of stress on voting is different for those who have already
established a habit. Because most voters form the habit of voting early in their voting eligibility (Franklin, 2004; Plutzer,
2002), we are merely attempting to identify those who have that habit or do not have that habit. Voting both in
the past U.S. presidential election and the mid-term congressional election (which regularly have much lower turnout
than presidential elections) suggests habit. Although we were not able to test it in the lab experiment, in the field
experiment we examined different operationalizations of habitual and nonhabitual voters ranging from having voted
in only the most recent election to having voted in all three previous general elections and found no significant differences
in effects.
10
Although we do not show the results here, we found no difference in the effects of the stress prime on the anticipated
turnout behaviors between individuals who voted in only one of the two elections and those who did not vote at all.
Differential Effects of Stress on Voter Turnout 541
models exclude individuals who were shown the brighten-your-life prime and use only those in the
control group and those who saw the prime prompting them to think about stressful things in their
day-to-day life. In addition to the experimental conditions and an individual’s past voting history, we
also have included a series of demographic and socioeconomic controls routinely found to influence
an individual’s participation rates.11
In our survey, we also asked individuals a series of questions to
measure their sense of political efficacy, and we include a summary measure of efficacy drawn from
those questions.12
The effect of priming stress is different for habitual and nonhabitual voters. As is seen in Table 3,
individuals who were primed to think about stressful life situations who voted in both the 2008 and
2010 elections estimated their likelihood of turning out just over 3 percentage points higher than habitual
voters who were not given the prime (increasing their self-perceived likelihood of voting from
94.7 to 97.7). On the contrary, those who had not voted in both elections and were exposed to the
same life-stress prime reduced their likelihood of voting by roughly 5 percentage points (decreasing
their perceived likelihood of voting from 68.8 to 63.9). Thus, as is evident from the coefficient on the
interaction in Table 2, introducing stress from daily-life interactions has an opposite effect among
habitual voters and nonhabitual voters. The interaction of stress with past voting experience produces
significantly different effects of stress among these different subgroups.
The second two models in Table 2 (and the last column in Table 3) also include the group that
was shown the additional prime asking respondents to reflect on things that made their life more pleasant.
These models demonstrate similar results. None of the additional variables or interactions are significant,
nor is there any change in the significance of the differential effect of focusing on stressful
day-to-day situations among habitual and nonhabitual voters. The effect sizes of stress and the interaction
of stress and habitual voter status in these models are similar to those in the first model. These
results show that the effects were the result of individuals being asked to think about stressful things
in their lives as opposed to the process of considering a topic more generally and thereby increasing a
respondents’ focus.13
However, the effect of asking individuals to think about things that brighten
their lives does not have the same differential effects on nonhabitual voters and habitual voters that
the life-stress prime does. Thus, we can rule out the possibility that the mere act of thinking about specific
facets of one’s life increases focus and alters the estimates of voting probability.
Table 3. Effect of Life-Stress and Brighten-Life Primes on Habitual and Nonhabitual Voters’ Likelihood of Voting
Compared to No Prime
No Prime Life Stress Brighten Life
Habitual Voter 94.7 97.7** 96.0
Nonhabitual Voter 68.8 63.9* 66.5
Note. One-tailed test with No Prime as the comparison group.
***p < 0.01, **p < 0.05, *p < 0.1.
11
In addition, we examined other variations and transformations of the age and income variables and found no effects.
The inclusion of these alternative measures does not alter the direction of the coefficients nor their significance.
12
Respondents were asked if they agreed, disagreed, or neither agreed nor disagreed with the following two statements:
“People like me don’t have any say about what the government does” and “Public officials don’t care much what people
like me think.” Individuals were given a score of 21 for each question to which they answered that they agreed
and a score of 1 for each question to which they disagreed. Questions to which they answered they neither agreed nor
disagreed were scored as 0. We then scaled the variable from 21 to 1.
13
In other models not shown here, we also included an interaction between an individual’s level of political efficacy
and the experimental prime they were shown. None of these interactions were significant nor did they have a substantive
effect on the other variables of interest, suggesting that it is not one’s belief in the effectiveness of the process,
but one’s familiarity with the process that is important.
542 Hassell and Settle
Field Experiment
There are large differences between asking about the likelihood of voting and measuring actual
participation. Individuals have an incentive to misrepresent socially acceptable behaviors such as voting
(Abramson & Claggett, 1984; Katosh & Traugott, 1981). An increase in habitual voters’ indication
of likelihood to vote from 94.7 and 97.7, or the decrease in the likelihood of voting of
nonhabitual voters from 68.8 to 63.9, may not represent a substantively meaningful difference in turnout.
For these reasons, we also conducted a field experiment using similar techniques to evaluate the
effect of priming life stress on actual voting behavior.
In this second experiment, we used a sample of 3,128 registered voters in Mount Vernon,
IA, to implement a single factor design with two levels.14
Mount Vernon is a small suburban
town located in the Cedar Rapids, IA, Metropolitan area. The town’s population at the 2010 census
was 4,506, of which 95% were White.15
The median income for the town was $70,960, with
roughly 11% of all families living below the federal poverty line. Prior to random assignment,
we removed the 672 individuals who listed an address associated with the small college in the
town.16
Random assignment to one of two groups was done at the individual level, with the
caveat that no more than one individual in a household could be part of the treatment group. We
eliminated multiple mailings to the same household because we were concerned about the multiplicative
effect of sending more than one mailer. In situations where multiple individuals were
randomly selected into the treatment group, we randomly moved all but one of those individuals
into the control group.
Larger field experiments have often focused on single-member households to avoid this
complication. However, we were concerned that there were insufficient single-member households
in this community to restrict our experiment in that way. To the degree that voter mobilization
may spread within a household (see Nickerson, 2008), it should be more difficult for us to
detect an effect between the treatment and control groups. However, our findings are also robust
Table 4. Demographics of Randomly Assigned Treatment and Control Groups
Receive Postcard Control
Age 47.00 48.59
Partisan 64.52% 63.52%
Republican 24.75% 23.96%
Democrat 39.06% 39.33%
Nonhabitual Voters 30.62% 32.44%
Did Not Vote in 2012 15.99% 14.02%
Total N 699 1757
Note. All differences between groups are not statistically significant except for the difference in age.
14
By excluding nonregistered individuals, our sample provides a conservative test of our hypothesis that there are differential
effects of stress on habitual and nonhabitual voters. If nonregistered voters have an even higher state of inertia,
if we find a differential effect between habitual voters and habitual nonvoters excluding these individuals, then it is
more likely that these differential effects would be stronger if we included nonregistered individuals.
15
We recognize that the racial makeup of the chosen city is unrepresentative of national population as a whole. Racial
minorities, however, are more likely to be nonhabitual voters, thus making it more difficult to find an effect among
that subgroup. Although some might argue that minorities may have developed adaptive strategies to deal with life
stressors (Pacheco & Plutzer, 2008), we looked and found no difference in the effects of stress on participation
between Whites and minorities in the lab experiment.
16
Because the election during which we ran the experiment was a local city election, we eliminated from the sample
those who were obviously students, as their participation was anticipated to be substantially lower than the general
population due to their temporary residence and decreased interest in local affairs.
Differential Effects of Stress on Voter Turnout 543
if we only look at single-member households in the sample. Hotelling balance tests reveal no significant
differences between the groups on partisanship as is shown in Table 4. There is a minor
difference in age; the treatment group is older on average by a mere two years, but there is no
significant difference between the ages of the habitual voters and the nonhabitual voters within
each treatment.
Individuals assigned to the treatment group were sent a 5.500
3 8.500
postcard that arrived in
their mailbox one to three days prior to the 2013 city council and mayoral election held on the
first Tuesday in November.17
Both the city council and mayoral elections were highly competitive
with the incumbents losing by margins of under 5%. The postcard asked subjects to fill out
and return a brief survey about the effects of stress in the local community. As with the first
experimental procedure, respondents were asked to list up to three things in their day-to-day life
that added stress to their lives. In addition, respondents were asked to also recall the last time
they experienced the stressor and to write a few sentences about how that recent stressful experience
made them feel. As before, we were careful to remove any reference to politics in the survey,
and the mailer did not mention anything at all about the election or the fact that political
Figure 2. Life-Stress mailer.
17
Previous research has provided some evidence that affective states can fade quickly—especially those induced by a
short stimulus (e.g., Mutz, 2015, pp. 86–87)—perhaps making it difficult to identify whether the mailer caused the
decrease in turnout. Random assignment of individuals to the treatment group should address this issue. However, we
recognize the typical trade-off between laboratory studies which are high on internal validity and low on external
validity, and field experiments which are high on external validity, but where it is often more difficult to identify the
causal mechanism (McDerrmott, 2011). It is for this exact purpose that we have done both a laboratory experiment
and a field experiment in order to help us identify the causal mechanism and also its real-world implications. Thus,
while it is possible that the effect of the mailer is more limited than we claim, both the laboratory and field experiments
point to the same result. In addition, previous work on the effect of mailings indicates that individuals are able
to identify and recall mail they received for at least five days and that these mailers can affect opinions and behaviors
for that same period of time and perhaps longer (Doherty & Adler, 2014). It is also important to remember that the
intent of the postcard was not to change the affective state, but rather to change individuals’ self-conception of themselves
by activating a biased set of memories that in turn leads them to conceptualize themselves as stressed individuals
(Kunda et al., 1993). Given that individuals are able to recall mail for at least a five-day period, we believe that
the effect of the postcard should last at least three days (the maximum amount of time between the arrival of the postcards
and Election Day), if not longer.
544 Hassell and Settle
scientists were conducting the study. Figure 2 provides a copy of the mailer sent out to registered
voters. We measure intent-to-treat effects because we have no way to measure the extent to
which our treatment was read by the treatment group, a common problem in get-out-the-vote
field experiments (Gerber & Green, 2012).18
Different Effects for Habitual and Nonhabitual Voters
Figure 3 shows the levels of turnout for individuals in the treatment and control groups
measured from official county voter records. Turnout citywide was 34.7%, but this figure
also includes the 672 college students eliminated from the sample. According to records
from the registrar of voters, the turnout among noncollege students was 43.6%. As in the
first experiment, there is no overall significant difference between individuals assigned to
the life-stress group and the control group. Although turnout among those in the treatment
group was 1.6% lower than turnout in the control group, the difference was not statistically
significant.
As before, the differential effect of the mailer is evident by looking at the effects of the treatment
on habitual voters and nonhabitual voters, visualized in Figure 3.19
We again measured habitual voters
as those individuals who had voted in both a presidential and congressional election, in this case the
Figure 3. Effects of treatment in field experiment, by voter status. Lines indicate 95% confidence intervals.
18
No postcards were returned as undeliverable; however, only a very small proportion of people (about 3%) actually
returned the “survey” to the drop site. However, we expect that simply reading the postcard may have been sufficient
to cause an effect.
19
Although few postcards were returned, only one of those returned listed two items and none listed only one. Thus,
while we are unable to systematically compare stress levels among habitual and nonhabitual voters, the consistent
responses to the postcard (and also to the Mechanical Turk survey) suggest that habitual and nonhabitual voters do
not vary in their ability to identify stress in their lives. Likewise, as before, we also attempt to control for covariates
to reduce the influence of baseline levels of stress.
Differential Effects of Stress on Voter Turnout 545
2010 and 2012 elections.20
While the effects of the treatment on habitual voters are miniscule and
insignificant, the treatment lowered turnout among nonhabitual voters by 6.1% compared to nonhabitual
voters who did not receive the postcard. The effect is even greater among those who did not vote
in the previous election. Registered voters who did not vote in the 2012 presidential election were
9.7% less likely to vote if they received the stress postcard than if they did not.21
While our field experiment methodology limits our ability to control for a multitude of other attitudinal,
demographical, and socioeconomic factors that may influence voter turnout, we can use data
available from the voter file to again model the effects of the postcard in a regression framework.
Table 5 contains a series of models predicting individual turnout. In addition to the experimental conditions
and an individual’s past voting history, we also have included a limited series of controls routinely
found to influence an individual’s participation rates.22
After including the controls, the effect of the treatment on nonhabitual voters is still substantial
and significant. Likewise, as before, there is no significant effect on the turnout rates of habitual
voters.
Table 5. Effect of Stress Inducing Postcard on Voter Turnout With and Without Controls
(1) Voted (2) Voted (3) Voted (4) Voted
Variables of Interest
Stress Treatment Group 20.027 20.064 0.046 20.077
(0.103) (0.110) (0.087) (0.097)
Nonhabitual Voter 21.762*** 21.427***
(0.147) (0.171)
Nonhabitual Voter 3 Stress Treatment 20.452** 20.527**
(0.246) (0.264)
Nonvoter 2012 21.756*** 21.227***
(0.193) (0.224)
Nonvoter 2012 3 Stress Treatment 21.110** 21.370**
(0.492) (0.527)
Demographic Controls
Registered as Partisan 0.116 0.216**
(0.108) (0.106)
Age 0.152*** 0.149***
(0.019) (0.019)
Age2
20.001*** 20.001***
(0.000) (0.000)
Constant 0.263*** 24.474*** 20.011 24.479***
(0.071) (0.479) (0.076) (0.472)
Observations 2,456 2,456 2,456 2,456
Pseudo R2
0.111 0.200 0.063 0.166
Log-Likelihood 21495 21346 21575 21402
Note. Logit coefficients with clustered standard errors in parentheses.
***p < 0.01, **p <0.05, *p < 0.1.
20
We also tested coding habitual voters as voters who had voted in all three of the previous general elections (as is consistent
with Franklin’s [2004] view of the formation of voting habit). We find no significant differences in effects.
Habitual voters (measured as having voted in all three previous elections) saw no significant decline in the likelihood
of turning out (47.6% turnout in the control compared to 46.7% turnout in the treatment group), while nonhabitual
voters were almost 10% less likely to turnout (15.3% turnout in the control group compared to 5.6% turnout in the
treatment group, significant at p < .05).
21
All of these finding are consistent and significant if we limit our analysis to single-voter households.
22
As before, we examined other variations and transformations of the age and found no difference in the effects. The
inclusion of these alternative measures does not alter the direction of the coefficients nor their significance.
546 Hassell and Settle
Discussion and Conclusion
Heightening voters’ awareness of the stressors in their lives has differential effects on voter turnout
depending on the degree to which people have participated in politics in the past. While there
appears to be a nominal increase in the self-reported likelihood that habitual voters will turn out to
vote as a result of stress, this finding was not replicated when actual voter turnout was measured.
However, our results for nonhabitual voters are consistent across two studies: Making stress salient
for those people least practiced in the political realm substantively and significantly reduces both their
intention of voting and their actual behavior.
These findings suggest that stress from life experiences may be a motivating factor in perpetuating
political inequalities. Those people who are more likely to experience higher levels of stress in
their lives are also the same individuals who are most likely to be nonhabitual voters (Palloni, 2005).
If those people most likely to be nonhabitual voters, and as such most likely to withdraw from the
political arena when they experience stress in life, are also those who are most likely to experience
high levels of stress in their lives on a regular basis, a vicious cycle of stress and detrimental stress
response may push these citizens further and further away from connecting their problems with political
solutions. This is concerning. These findings suggest that stress from individuals’ life experiences
may be a motivating factor in perpetuating current political inequalities between the haves and havenots
in society.
Although our study cannot disambiguate the mechanism of the effect of stress, we suspect that it
has to do in part with differences in the familiarity with the political process between habitual voters
and nonhabitual voters. Unlike nonhabitual voters, habitual voters are more likely to recognize connections
between life stress and the electoral process. Habitual voters are less likely to consider voting
an arduous task and a burden. Nonhabitual voters, however, likely perceive voting as a costly demand.
They are more likely to be unfamiliar with the process and are less able to identify clear connections
between their electoral participation and the stress in their lives. For nonhabitual voters, the act of voting
during times of increased stress becomes only an additional, costly task adding to, instead of
reducing, the level of stress in their lives.
All of this is also consistent with Blais’s (2000) finding that those who have never voted tend to
overestimate the costs associated with voting. People who have never voted perceive that the act of
voting takes more time than do people who have voted previously, and a lower percentage of nonvoters
report that voting is “easy.” Thus, for nonhabitual voters, the act voting is seen as both more
costly and less beneficial and is more likely to be set aside in the face of life’s other problems.
One potential alternative explanation for our results is that the mechanism of the effect for the
field experiment was a form of the Hawthorne Effect, where people are more likely to adhere to
socially desirable behavior when they know they are being watched. We do not think this explains our
results for two reasons. First, while the treatment postcard did mention one of the researcher’s names
and affiliation with a local college, it did not indicate that he was a political scientist. The subject
would have had to independently seek out this information on the Internet or in the college’s faculty
directory. Furthermore, the postcard mentioned nothing about politics or the municipal election, so
there was no easy way for a subject to know what the desirable behavior was. Second, our results indicate
a pattern that runs counter to what one would expect if subjects were conforming to the normatively
desirable behavior of voting. Habitual voters—who should be most aware of and sensitive to
that norm—were unaffected by the postcard, while nonhabitual voters were less likely to turn out.
The vast majority of voter mobilization mailings sent out makes explicit mention of the election
at hand or somehow makes voting salient to the recipient. This makes sense: Campaigns, interest
groups, and academics have an interest in helping people channel their behavior toward participation
in the electoral sphere. Thus, on the one hand, our treatment is unrealistic as far as a representation of
the kind of political mail that people might actually receive. However, people are exposed to triggers
Differential Effects of Stress on Voter Turnout 547
for stress in every facet of their lives. Whether it is the arrival of an unpaid bill or credit card statement
or a conversation with a friend about personal problems, life stress is made salient day in and day out.
In fact, those with the greatest amount of stress are least likely to be able to avoid it.
Our results suggest that factors outside of the specific realm of politics can have an important
effect on the decisions that people make within the political sphere and that people’s sensitivity to
stress and the way they learn to manage it may have important repercussions on levels of political participation
in our society. As a result, we hope that as policy makers consider how to encourage voting
and work to reform the voting process to include those marginalized from the democratic process,
they will be more concerned with ways to reduce life stress among those who do not participate regularly.
By identifying times where stress is lowest (Ingraham, 2014; M€akinen & Kinnunen, 1986), policy
makers may be able to encourage turnout among those most marginalized in society. Otherwise,
stress in life will continue, as it has in the past, to perpetuate political inequalities.
ACKNOWLEDGMENTS
We are grateful for the comments and suggestions provided by Elaine Denny, Melinda Green,
Julie Pacheco, Johanna Schuster-Craig, Dane Wendell, and Rebecca Wines, as well as those provided
by three anonymous reviewers and the editor of the journal. We also want to acknowledge the
assistance of Drew Engelhardt with the data collection, Alex McGee with the coding of stress content,
and Meg Schwenzfeier with figure preparation. Lastly, we are thankful to Jake Krob, Rich
Eskelsen, and the rest of the staff at the Mount Vernon-Lisbon (Iowa) Sun for their help in the project.
All mistakes, of course, remain our own. Correspondence concerning this article should be
addressed to Hans J. G. Hassell, Cornell College, 600 First St. SW, Mount Vernon, IA 52314. Email:
hhassell@cornellcollege.edu
REFERENCES
Abraham, K. G., Maitland, A., & Bianchi, S. M. (2006). Nonresponse in the American time use survey: Who is missing
from the data and how much does it matter? Public Opinion Quarterly, 70(5), 676–703.
Abramson, P. R., & Claggett, W. (1984). Race-related differences in self-reported and validated turnout. Journal of Politics,
46(3), 719–738.
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84(2), 191–
215.
Berinsky, A. J. (2004). Silent voices: Public opinion and political participation. Princeton, NJ: Princeton University
Press.
Berinsky, A. J., Huber, G. A., & Lenz, G. S. (2012). Evaluating online labor markets for experimental research: Amazon.com’s
Mechanical Turk. Political Analysis, 20(3), 351–368.
Blais, A. (2000). To vote or not to vote? The merits and limits of rational choice. Pittsburgh, PA: University of Pittsburgh
Press.
Brady, H. E., Verba, S., & Schlozman, K. L. (1995). Beyond SES: A resource model of political participation. American
Political Science Review, 89(2), 271–294.
Brody, R. A., & Sniderman, P. M. (1977). From life space to polling place: The relevance of personal concerns for voting
behavior. British Journal of Political Science, 7(3), 337–360.
Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s Mechanical Turk: A new source of inexpensive, yet
high-quality data? Perspectives on Psychological Science, 6(1), 3–5.
Coppock, A., & Green, D. (2015). Is voting habit forming? New evidence from experiments and regression discontinuities.
American Journal of Poltiical Science. doi:10.1111/ajps.12210
Doherty, D., & Adler, E. S. (2014). The persuasive effects of partisan campaign mailers. Political Research Quarterly,
67(3), 562–573.
Downs, A. (1957). An economic theory of democracy. New York, NY: Harper and Row.
548 Hassell and Settle
Duffy, J., & Tavits, M. (2008). Beliefs and voting decisions: A test of the pivotal voter model. American Journal of
Political Science, 52(3), 603–618.
Druckman, J. N., & Kam, C. D. (2011). Students as experimental participants: A defense of the “narrow data base.” In
J. N. Druckman, D. P. Green, J. H. Kuklinski, & A. Lupia (Eds.), Handbook of Experimental Political Science (pp.
41–57). New York, NY: Cambridge University Press.
Fieldhouse, E., & Cutts, D. (2012). The companion effect: Household and local context and the turnout of young people.
Journal of Politics, 74(3), 856–869
Franklin, M. N. (2004). Voter turnout and the dynamics of electoral competition in established democracies since 1945.
New York, NY: Cambridge University Press.
Gerber, A. S., & Green, D. P. (2012). Field experiments: Design, analysis, and interpretation. New York, NY: W.W.
Norton.
Gerber, A. S., Green, D. P., & Shachar, R. (2003). Voting may be habit-forming: Evidence from a randomized field
experiment. American Journal of Political Science, 47(3), 540–550.
Grant, J. T., & Rudolph, T. J. (2002). To give or not to give: Modeling individuals’ contribution decisions. Political
Behavior, 24(1), 31–54.
Green, D. P., & Shachar, R. (2000). Habit-formation and political behaviour: Evidence of consuetude in voter turnout.
British Journal of Political Science, 30(4), 561–573.
Grimmer, J., Messing, S., & Westwood, S. J. (2012). How words and money cultivate a personal vote: The effect of legislator
credit claiming on constituent credit allocation. American Political Science Review, 106(4), 703–719.
Hassell, H. J. G., & Monson, J. Q. (2014). Campaign targets and messages in campaign fundraising. Political Behavior,
36(2), 359–376.
Hassell, H. J. G., & Visalvanich, N. (2015). Call to (in)action: The effects of racial priming on grassroots mobilization.
Political Behavior, 37(4), 911–932.
Highton, B. (2004). Voter in the registration and turnout in the United States. Perspectives on Politics, 2(3), 507–515.
Hobbs, W. R., Christakis, N. A., & Fowler, J. H. (2013). Widowhood effects in voter participation. American Journal of
Political Science, 58(1), 1–16.
Ingraham, C. (2014). The Google Misery Index: The times of year we’re most depressed, anxious and stressed. Washington
Post. December 3.
Katosh, J. P., & Traugott, M. W. (1981). The consequences of validated and self-reported voting measures. Public Opinion
Quarterly, 45(4), 519–535.
Krupnikov, Y., & Levine, A. S. (2014). Cross-sample comparisons and external validity. Journal of Experimental Political
Science, 1(1), 59–80.
Kunda, Z., Fong, G. T., Sanitioso, R., & Reber, E. (1993). Directional questions direct self-conceptions. Journal of
Experimental Social Psychology, 29(1), 63–86.
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York, NY: Springer.
Levine, A. S. (2015). American insecurity: Why our economic fears lead to political inaction. Princeton, NJ: Princeton
University Press.
M€akinen, R., & Kinnunen, U. (1986). Teacher stress over a school year. Scandinavian Journal of Educational Research,
30(2), 55–70.
Mattila, M., S€oderlund, P., Wass, H., & Lauri, R. (2013). Healthy voting: The effect of self-reported health on turnout in
30 countries. Electoral Studies, 32(4), 886–891.
McDermott, R. (2011). Internal and external validity. In J. N. Druckman, D. P. Green, J. H. Kuklinski, & A. Lupia
(Eds.), Handbook of experimental political science (pp. 27–40). New York, NY: Cambridge University Press.
McGuire, W. J., & McGuire, C. V. (1996). Enhancing self-esteem by directed-thinking tasks: Cognitive and affective
positivity asymmetries. Journal of Personality and Social Psychology, 70(6), 1117–1125.
Milbrath, L. W. (1965). Political participation: How and why do people get involved in politics. Chicago, IL: Rand
McNally.
Mondak, J. J. (2008). A framework for the study of personality and political behaviour. British Journal of Political Science,
38, 335–362.
Mondak, J. J., Hibbing, M. V., Canache, D., Seligson, M. A., & Anderson, M. R. (2010). Personality and civic engagement:
An integrative framework for the study of trait effects on political behavior. American Political Science
Review, 104(1), 85–110.
Mutz, D. (2015). In your face politics: The consequences of uncivil media. Princeton, NJ: Princeton University Press.
Ojeda, C. (in press). Mental disability and political voice: The effect of depression on political participation. Social Science
Quarterly.
Differential Effects of Stress on Voter Turnout 549
Pacheco, J., & Fletcher, J. (2015). Incorporating health into studies of political behavior: Evidence that health impacts
turnout and partisanship. Political Research Quarterly, 68(1), 104–116.
Pacheco, J., & Plutzer, E. (2007). Stay in school, don’t become a parent: Teen life transitions and cumulative disadvantages
for voter turnout. American Politics Research, 35(1), 32–56.
Pacheco, J., & Plutzer, E. (2008). Political participation and cumulative disadvantage: The impact of economic and social
hardship on young citizens. Journal of Social Issues, 64(3), 571–593.
Palloni, A. (2005). Reproducing inequalities: Luck, wallets, and the enduring effects of childhood health. Demography,
43(4), 587–615
Peterson, S. A. (1987). Biosocial predictors of older Americans’ political participation. Politics and the Life Sciences,
5(2), 246–251.
Peterson, S. A., & Gabbidon, S. L. (2007). Stressful life events and African American political participation: A statelevel
analysis. Journal of Black Studies, 38(2), 142–154.
Peterson, S. A., & Somit, A. (1994). Bad times and the political behavior of older African Americans. American Review
of Politics, 15(1), 43–56.
Plutzer, E. (2002). Becoming a habitual voter: Inertia, resources, and growth in young adulthood. American Political Science
Review, 96(1), 41–56.
Powell, G. B. (1986). American voter turnout in comparative perspective. American Political Science Review, 80(1), 17–
43.
Rosenstone, S. J., & Hansen, J. M. (1993). Mobilization, participation, and American democracy. New York, NY:
Macmillan.
Sandell, J., & Plutzer, E. (2005). Families, divorce and voter turnout in the US. Political Behavior, 27(2), 133–162.
Schur, L., Shields, T., Kruse, D., & Schriner, K. (2002). Enabling democracy: Disability and voter turnout. Political
Research Quarterly, 55(1), 167–190.
Sniderman, P. M., & Brody, R. A. (1977). Coping: The ethic of self-reliance. American Journal of Political Science,
21(3), 501–521
Verba, S., Schlozman, K. L., & Brady, H. E. (1995). Voice and equality: Civic voluntarism in American politics. Cambridge,
MA: Harvard University Press.
Wolfinger, R. E., & Rosenstone, S. J. (1980). Who votes? New Haven, CT: Yale University Press.
Weinberg, J., Freese, J., & McElhattan, D. (2014). Comparing data characteristics and results of an online factorial survey
between a population-based and a crowdsource recruited sample. Sociological Science, 1(August), 292–310.
Supporting Information
Additional supporting information may be found in the online version of this article at the publisher’s
website:
Analysis of the Sources of Stress for Habitual and Nonhabitual Voters
Table A1. Testing for Content Differences in the Stressors Listed by Habitual Voters and Nonhabitual
Voters.
550 Hassell and Settle