Content uploaded by Caitlyn Bradfeild
Author content
All content in this area was uploaded by Caitlyn Bradfeild on May 18, 2016
Content may be subject to copyright.
Bradfield Johnson 1
The Effect of Making Election Day a Holiday
An Original Survey and a Case Study of French Presidential
Elections Applied to the U.S. Voting System
Caitlyn Bradfield
1
Paul Johnson
2
April 18, 2016
Abstract
In order to better understand the effect that designating Election Day as a national holiday has on
voter turnout, we conducted a case study comparing France and the United States. Average turnout
among developed democracies averages 80% in presidential elections (Powell 1986, 17).
However, voter turnout in the United States is one of the lowest among developed democracies;
only 75.1% of registered American voters turned out to vote between 2002 and 2012. We
conducted a case study comparing France and the United States, two countries that are
institutionally similar, in an effort to predict how voter turnout in the United States would change
if Election Day was designated as a national holiday. We hypothesized that by so doing, voter
turnout in America would increase and even surpass turnout levels evident in France. In addition,
we conducted an original survey and found that creating a national holiday for Election Day would
increase voter turnout by about 16 percentage points. This significant increase would cause voter
turnout in the U.S. to be consistent with other developed democracies. We argue that it is vital to
the American political system to minimize barriers to voting and believe that maximizing voter
turnout is best achievable through the creation of a national Election Day holiday.
1
Political Science Undergraduate, Brigham Young University. Email: caitlyn.bradfield@gmail.com
2
Political Science Undergraduate, Brigham Young University. Email: paul@paulj.us
Bradfield Johnson 2
Introduction
Voter turnout in the United States of America has lagged behind other developed democracies for
decades. Exactly what causes this discrepancy has been an issue of debate. Ironically, “voters [in
the United States] are more politically aware and involved than citizens in any other democracy,
yet the levels of voter turnout are consistently far below the democratic average” (Powell 1986,
17).
We theorize that voter turnout is largely dependent on the ability and relative ease that
voters have to vote. In developed democracies, eligible voter turnout averages around 80% in
general elections (Ibid., 17). However, elections in the United States of America present a different
story as turnout is much lower, often averaging around 50% of all eligible adults (Bipartisan Policy
Center 2012) and 75% of all registered, eligible adults (IDEA 2016). In the 2012 election,
registered voter turnout reached a low of 67% of eligible adults. In order to maximize voter turnout,
we theorize that designating Election Day in the U.S. as a national holiday will rapidly increase
voter turnout in the U.S. by minimizing possible obstacles that voters face on Election Day. We
hypothesize that most voters fail to vote due to time and responsibility constraints. To predict the
effect that introducing a national election holiday would have on voter turnout, we conducted a
case study comparing two relatively similar democracies: The United States of America and the
French Republic.
According to our calculations, voter turnout among registered American voters averaged
75.16% of all registered voters between 2002 and 2012
3
. Existing research suggests that turnout
in America is encouraged by a unique set of “political attitudes” that, if possessed by other
democracies, would increase their turnout (Powell 1986, 18). However, while this benefit exists,
3
See Table A1
Bradfield Johnson 3
“institutional factors” and difficult registration laws dramatically reduce turnout among American
voters (Ibid.). Indeed, additional research suggests that voter turnout among democracies is a result
of political institutions and electoral law (Jackman 1987, 405). While many countries schedule
Election Day on a weekend, or create a national holiday for elections, the United States has
designated the Tuesday after the first Monday in November as Election Day.
We hypothesize that the day on which general elections are held in the United States serves
as a barrier and reduces voter turnout among working class Americans who are already registered
to vote. While polls are generally open from 7:00am - 7:00pm, many working adults who are
eligible to vote face long commutes to work or may work longer hours than the average American.
In addition, voters with families may struggle to find time to wait in line. Reports indicate that
average wait times vary dramatically state-to-state, with wait times averaging two minutes in
Vermont and nearly 40 minutes in Florida (CCES 2012, SPAE 2012). It is estimated that between
500,000 - 700,000 votes are lost due to voters being deterred by long lines (Stewart and
Ansolabehere 2013). Most of these long lines are caused by workers getting in line to vote before
or after work. We believe wait times would be reduced with the implementation of our hypothesis
as it would open up more hours of the day in which workers could potentially vote.
Literature Review
According to the U.S. Census Bureau in 2014, individuals cited a variety of reasons for failing to
vote. Nearly 30% of voters who failed to vote in 2014 cited being “too busy” to cast their vote,
followed by 16% being “not interested”. In that same year, 11% of voters did not vote due to illness
whereas 10% did not vote due to being out of town. Among the many other reasons to skip an
election were: “forgot to vote,” “disliked candidates/issues,” “registration problems,”
“inconvenient polling place,” and “transportation issues” (U.S. Census Bureau 2014). We theorize
Bradfield Johnson 4
that most of these reasons could be reasonably resolved with the implementation of our hypothesis
and will hereafter explore in more depth why we believe a national holiday for election day would
not increase the number of voters who miss due to being out of town.
Anthony Downs, in his influential research about the costs of voting, found that “where
voting is costly, individuals will consider both how much they care about the outcome and the
likelihood that their vote will influence the outcome (be pivotal).” Downs’ theory was later
developed into a model, which outlines expected behavior by voters, as “R = (B * P) - C + D,
where
➔ B is the utility gain from getting the preferred outcome,
➔ P is the probability that the individual’s vote will yield the preferred outcome (the
probability that the individual is pivotal)
➔ C is the (non-negative) cost of voting, and
➔ D is the positive benefit of the act of voting” (Farber 2009).
Indeed, conventional wisdom suggests inherent costs in voting. Voting is time intensive
and may conflict with other “demands and preferences” (Stein and Vonnahme 2008, 487). For
working Americans, poll voting times may be inconvenient, especially in cases where commutes
to work are required (Gimpel and Schuknecht 2003, Dyck and Gimpel 2005, Gimpel, Dyck, and
Shaw 2006). These working Americans with commutes often spend hours in slow traffic on their
way to work and find it difficult to vote as most polls are open between 7:00am and 7:00pm. These
voters, while potentially politically involved, face a conflict of interest between work requirements
and civic responsibilities and are more likely to vote early as a result (Gronke College 2004).
Despite this, a significant portion (25%) of Americans making over $150,000 per year fail to vote
due to time constraints.
Bradfield Johnson 5
In addition, voters who must wait for extended periods of time in order to vote are
dramatically less likely to vote and it is predicted that in between 500,000-700,000 votes are lost
due to lines at polls (Stewart and Ansolabehere 2013). The number of votes lost due to avoidable
situations is often a politicized debate. In the 2016 Arizona democratic primary, voters in Maricopa
County were forced to wait hours in line to vote since Maricopa County “only had one polling site
for every 21,000 voters” (Rayman, O’Dell, and Cano 2016). Many voters are unable to devote
large amounts of time per day to spend at polls and are forced away by long lines when other
priorities require attention. We argue that resolving the many issues surrounding voting in the
United States is an ethical issue, and thus, of utmost importance.
As such, it is sufficiently evident that voters are deterred for a variety of reasons. Most
prevalent among all, however, are reasons relating to time management and inconvenience. As has
been illustrated, voters in America often face a conflict of interest on Election Day between
managing important personal affairs and their civic responsibilities.
Countries apply a variety of approaches in order to combat low voter turnout rates. In
Australia, for instance, voter turnout nears 100%. It is worth noting that Australia’s Election Day
occurs on a Saturday and is mandatory. While we recognize that either, or both, institutions could
be the driving factor for high turnout, the focus here is that other democracies around the world
are trying to address voter turnout problems. We are not recommending mandatory voting laws,
but find it important to learn from other democracies that there are ways to increase turnout as
research shows that mandatory turnout laws are “not necessary for high levels of participation.”
(Hirczy 1994).
Evidence of this is visible in France. Turnout of registered voters in France averaged
81.34% in the years 2002-2012 without mandatory voting laws. France schedules elections on
Bradfield Johnson 6
weekends in to minimize the possibility that voters may be unable to vote due to various time and
responsibility constraints.
France and Australia are not alone in their efforts. Germany, New Zealand, and other
countries either vote on a weekend or designate Election Day as a national holiday. While many
countries vote on a weekend, we hypothesize that establishing Election Day as a national holiday
will be more beneficial to voter turnout as it would solve many of the issues that voters cite for
failing to vote without requiring a constitutional amendment. Martin Wattenberg succinctly
summarized the benefit that this would have towards Americans when he said that, “this would
send a strong signal about the importance our country attaches to voting.”
Case Study: France
Why France?
The idea that making Election Day a national holiday would increase registered voter turnout is
the result of a case study comparing the United States and France. We believe that the only
significant difference between these countries in relation to voter turnout is the day of the week on
which Election Day is held. This is the principal reason we have chosen to compare France and
the United States. Because there is no work or other weekday responsibilities, voters turn out at
higher rates, averaging 81% over the last three general elections.
4
We believe that a national
holiday Election Day would simulate this type of result in the United States and even increase it
as people are less likely to be out of town on a Tuesday than on a Sunday.
4
See Table A1
Bradfield Johnson 7
France and the United States as Most Similar Cases
We intend to prove that France and the United States are similar except for a particularly critical
difference: Election Day. France and the United States have been used as most similar cases in
previous research on governmental structure and political involvement, thus laying the foundation
for a study on elections (Holtz-Bacha, Kaid, and Johnston 2010, Conley 2007). In literature by
Holtz-Bacha, Kaid, and Johnston, it was concluded that “there are definite shared cultural and
political values that give rise to similar electoral strategies” between the United States and France
(2010).
We found that France and the United States are comparable in terms of government
structure. Previous research shows that “France and the United States are the only two countries
that have an unambiguously presidential form of government” (Jackman 1987, 413 Lihphart 1984,
70-71). Institutionally, the two countries are both republics and have, judiciary, executive, and a
bicameral legislative branch of government. While registration in the United States is not
automatic as it is in France, due to the National Voting Act of 1993, citizens are now able register
to vote when they obtain their licenses, minimizing this difference. Additionally, citizens in both
countries may register to vote at the same age: 18 (Jackman 1987, 414).
Both countries have executive branches independent of the parliamentary branch, unlike
other comparable countries (Jackman 1987, 413, Lihphart 1984, 70-71). The term lengths for the
president of France and the United States are similar, at five and four years respectively. In fact,
the executives in the United States are France are nearly identical, even down to the veto powers
and way the cabinet systems operate (Conley 2007). The only significant executive difference is
the fact that a dual-executive exists in the French institutional structure. Additionally, Conley
determined that the countries are similar enough to compare in terms of political activity, thus we
Bradfield Johnson 8
believe that the countries must be similar enough to compare with a specific type of political
activity: voting.
Economically, the two countries are also relatively similar. GDP per capita in the U.S.
averages is $54,000, whereas GDP per capita in France is $43,000 (Find The Data 2016).
Theoretically, larger countries have a higher capacity for a higher GDP per capita. In a study by
the Organization for Economic Cooperation and Development, it was found that “Returns to
physical and human capital might be higher in countries having a better access to large markets”
(Boulhol, de Serres, and Molnar, 2016, 7). Considering the dramatic difference in geographic size
between the countries, we find this difference to be negligible.
France and the United States share many demographic similarities. The median age in
France and the United States is only separated by 3.3 years and the average sex ratio is only
separated by .01 males per females (Ibid.). In France, 21% of the population lives in rural settings
whereas 18.7% in the U.S. live in similar settings (Ibid.). Just over 81% of Americans live in Urban
centers while nearly 80% of French citizens live in Urban centers (Ibid.). Birth rate and death rate
in the two countries is nearly identical and the educational system are very similar (Ibid.).
While the two countries are inherently different, we find evidence to suggest that the
demographics and political institutions are comparable, as do scholars in previous studies. These
similarities will hereafter be discussed in greater detail.
Voting procedures in France mandate that elections be held on Sunday in order to
maximize voter turnout. France averaged about 81.34% turnout in between 2002-2012, which is
comparable to averages for other developed democracies (IDEA). While our theory calls for an
Election Day national holiday, we justify that France’s weekend Election Day is comparable since
both provide traditionally non-work days for voters to cast ballots. We expect that a national
Bradfield Johnson 9
election day holiday will increase voter turnout to levels that are consistent with and perhaps even
higher than with other developed democracies as more individuals will be granted time off.
We chose to compare voter turnout using turnout percentage of registered voters over the
space of a nine-year period, 2002 through 2010. Using percentage of registered voters instead of
the traditional VAP (voting age population) measure eliminates “not registered” as a reason that
people do not vote and allows us to focus on day-of inconveniences. In addition, it minimizes the
disproportional effect that immigration may have on voter turnout. This is particularly important
today as France and the United States have large numbers of non-eligible adults (either due to their
criminal or immigration status), which may distort the VAP ratings. We used data from IDEA
5
to
examine the executive election turnout in both France and the United States in each election
occurring in this time frame.
Over the last decade, the United States has seen a decrease in voter turnout. Interestingly
though, the 2004 election saw a greater turnout in the United States than the 2002 election in
France. We suspect that this was a result of great American patriotism that followed the 9/11 terror
attacks and the war in Iraq. However, even before 2004, voter participation was greater than in
recent years, according to IDEA. We suspect that the National Voting Act of 1993 influenced the
number of eligible voters who became registered voters, at a lagged rate, since voter registration
became as simple as checking a box at the DMV. France has automatic voter registration at age
18, thus the registered voting age population is very high. We believe that these countries are
increasingly comparable as the United States’ National Voting Act of 1993 increased the
percentage of the voting age population that are registered to vote. Of note is the idea that
5
See table A1 for results
Bradfield Johnson 10
registration is voluntary in the United States. Theoretically, voter turnout in the United States
among registered voters should even be higher than turnout in France since American citizens must
be politically interested enough to register.
We hypothesize that the large difference in turnout in executive election is a result of the
day of week that each country chooses to hold elections on. In France, elections are held on Sunday
and in the United States they are held on Tuesday. We believe that Sunday elections make voting
more accessible to the general population than a midweek election would. Moving Election Day
to a Sunday in the United States would require a constitutional amendment, which is not the most
feasible of plans. We propose instead that an election day holiday be instituted as it is similar to
holding elections on a weekend and would be more likely to be instituted by congress.
There are several additional advantages to making Election Day a holiday over moving it
to a weekend. First, Tuesday has been Election Day in the United States since 1845 and has become
an ingrained part of American tradition. Election Day was designated on a Tuesday in because it
“did not interfere with the Biblical Sabbath or market day, which was on Wednesday in many
towns” (Huffstutter 2006). Second, a Tuesday Election Day minimizes the number of people that
would fail to vote due to travels, as they would need to be in town for work on Monday and
Wednesday. Failing to vote is easier if Election Day is on a weekend when many voters have time
off and choose to travel. We found that 19% of those that did not vote in a recent election failed to
do so because they were “out of town.” This number would only increase if we gave greater ability
for travel in Election Day. Designating Monday as Election Day may encourage some voters to
travel out of town for a three-day weekend, whereas a Tuesday Election Day makes it harder to
miss multiple days of work to travel. While a holiday Election Day would fail to solve the problem
Bradfield Johnson 11
of many workers who travel for extended periods of time for business, it would not increase the
likelihood of travel among other citizens significantly.
Methodology
Survey: Why Registered Voters Fail to Vote6
To provide quantitative support for our hypothesis, we conducted a case study administered
through Brigham Young University’s Research Project in American Politics class. The survey
included a variety of questions from students performing research. Our specific questions will
address three topics: reasons an individual did not vote in previous elections, useful methods to
make voting easier, and an opinion question on making Election Day a national holiday. The
survey questions follow:
6
Our study focused on turnout of registered voters in the United States and in France in order to minimize the
disproportional and negative impact that non-registered adults may have on voter turnout. We eliminated
respondents from the survey that were not registered to vote.
Bradfield Johnson 12
Figure 1
7
,
8
Figure 2
7
We grouped “out of town” and not a registered voter” into the same category because both of these reasons cannot
be solved by making Election Day a national holiday. We limited the respondents’ options to prevent them from
becoming bored with or tired of the survey so some of the categories were groups together with other “like”
categories.
8
In order to ensure respondent honesty, we asked an additional question to survey respondents. The question reads
similarly to that in Figure 1, but this time asking specifically why respondents thought their neighbors failed to vote
in a recent election. The question reads, “There are a variety of reasons people do not vote in Presidential elections.
In fact, only 53% of eligible voters made it out to the 2012 Presidential election. What are some reasons your
neighbors are unable to make it to a polling location on election day? (Rank: 1 being biggest reason my neighbors
are unable to make it to a polling place).” We added this additional question to gauge whether or not respondents
answered accurately the question in Figure 1 honestly under the assumption that people answer for their neighbors
as they would themselves. Further, we omitted an “I always vote” option to ensure honest answers and avoid any
unintended biases. We found no statistical difference in any of the answers except “other.” See Figure A5 for visual.
Bradfield Johnson 13
Figure 3
Respondents were also asked whether or not making Election Day a national holiday would
solve the main reason that kept them from voting in a recent election and a variety of demographic
questions. This survey was administered via MTurk.
We hypothesized that these questions would give us information that would show that
voters fail to vote primarily due to inconveniences such as commuting or work hours. Just as
Sunday voting removes most inconveniences from voting, we hypothesize that a national holiday
would have a similar effect.
These questions, in addition to demographic questions allowed us to run a logit model in
order to predict what effect making Election Day a national holiday would have on those who did
not vote for a variety of reasons. The dependent variable in question is, “Would making Election
Day a national holiday solve the problem of x” and the independent variable of interest is the
reason people failed to vote.
Bradfield Johnson 14
Results
To analyze the survey responses we ran a logit model
9
. The model contained the variables of
interest (work/too busy, familial demands, not enough time/long lines, and forgetting to vote)
ranked as a first response. The model shows the predictive margins and the resulting percent
change. Using this model, we also analyzed specific groups of respondents by race, income,
ideology, and by whether they lived in a rural area.
We find that respondents who cited the variables of interest as the number one reason they
do not vote would be able to vote by making Election Day a national holiday.
9
See Table 1
Bradfield Johnson 15
Table 2
Logit Model: Would making an Election Day a Holiday solve the problem?
Pseudo R2 = .223
Variable
Coef
Std Er
Work/too busy (First Response)
2.57*
.220
Family demands (First Response)
1.58*
.358
Not enough time/long lines (First Response)
2.25*
.355
Forgetting to vote (First Response)
1.67*
.314
Predictive Margins
Predicted Probability
Change
Work/too busy (First Response) = no
.35
Work/too busy (First Response) = yes
.85
= 50%
Family demands (First Response) = yes
.52
Family demands (First Response) = no
.78
= 26%
Not enough time/long lines (First Response) = yes
.50
Not enough time/long lines (First Response) = no
.86
= 36%
Forgetting to vote (First Response) = yes
.51
Forgetting to vote (First Response) = no
.78
= 27%
* Significant at the .01 level. Variables held at their means. Predicted margins indicate the change that would be seen from moving from a 0
to a 1 on the binary scale, i.e. when work/too busy is the reason that the respondent did not vote, introducing a national holiday for election
day increases turnout of those respondents 50%.
This logit was run with controls. To see the full output see Table A2 in the appendix.
Bradfield Johnson 16
Discussion
The results that we found gave credence to our hypothesis that voter turnout would be increased
through eliminating barriers to voting in a way similar to what France has done. In addition, the
quantitative results of our survey corroborate our hypothesis.
Main Findings
We uncovered several astonishing statistics through our survey. The graph below visually displays
the general predicted effect of a making Election Day a national holiday on voter turnout among
those who ranked one of the variables of interest as first. We would see significant increases in
voter turnout among those respondents that cited our variables of interest as the primary reason
they fail to vote.
Figure 4
For individuals who failed to vote in a previous election due to work or being too busy, we
found that creating a national holiday for Election Day would increase voter turnout by 50%. For
those that did not vote due to familial demands, we expect voter turnout to increase by 26%. A
Bradfield Johnson 17
significant portion of respondents did not vote due to not having enough time or lines being too
long. We estimate that voter turnout for these individuals will increase by 36%. Lastly, we found
that voter turnout will increase by 27% for those who “forgot to vote” in a recent election. With at
least a 25% increase per variable, these findings are significant. All of these findings were
consistent with data derived from the 2010 Census.
A brief analysis of the data set suggests that about 50% of our sample selected one of our
variables of interest (work/too busy, familial demands, not enough time/long lines, and forgetting
to vote) as the first or second reason they were likely to have failed to vote. Extrapolated to the
general population, if we were to address the problem that 50% of non-voting Americans faced in
2012, we would have seen an increase in turnout by registered voters from about 67% to about
83%. That translates to just under 32,000,000 additional votes. This increase would put American
voter turnout on par with, and even above, the average turnout rate for most developed
democracies.
We understand that not every person in that group would choose to vote, but the potential
alone of adding 32,000,000 votes to the election is enough to justify our findings. If only half of
those potential votes were cast (which is reasonable considering registered voter turnout was at
67% in 2012) that is still an increase of 16,000,000 votes.
The Effect on Subgroups
We explored different subgroups of the population that tend to have lower voter turnout and the
reasons that they cite (U.S. Census Bureau 2014). We ran the logit model and included a condition
at the end that would reflect the subgroup of interest, then ran the predictive margins, for each
subgroup. The results of the predictive margins are below in Table 2. We chose each of these
subgroups for theoretical reasons that we will explain further below.
Bradfield Johnson 18
Table 3
Break-down by Subgroup
Race
Reason failed to vote
= 0
= 1
Expected change (in percent increase)
Black
Work/too busy
.50
.70
20%
Familial demands
.57
.88
31%
Not enough time/long lines
.53
.87
34%
Forgetting to vote
.57
.74
17%
Income
<$10,000
Work/too busy
.30
.80
50%
$100,000-$250,000
Work/too busy
.37
.92
55%
Ideology
Moderate
Work/too busy
.27
.79
52%
Not enough time/long lines
.37
.75
38%
Forgetting to vote
.38
.83
45%
Rural/Urban
Rural
Not enough time/long lines
.43
.92
49%
“=0” and “=1” indicate the predicted margins for each subgroup and for the specified reason the respondent failed to vote.
Black Respondents
Based on census data and established literature we recognize that racial minorities turn out at lower
rates than than White voters. We found that in each of the relevant reasons respondents fail to vote
that turnout among black voters may be increased by up to 34 %.
We suspect that the actual effect is higher. Many people can easily cite multiple reasons
they do not vote, so solving only the “not enough time/long lines” problem may not have elicited
a positive response for our dependent variable. Since many people selected one of the variables of
interest as both the first and second reason they do not vote we believe that instituting an Election
Day national holiday would solve many of the problems that black voters may face.
We also recognize that it is possible black citizens just choose to not vote at comparative
levels, a problem our solution cannot solve. Even if this is the case, 34% of respondents claimed a
Bradfield Johnson 19
national holiday would give them incentive to turnout when they fail to vote due to not enough
time or long lines. We hypothesize that this would substantially impact the percentage of registered
black voters that make it to the polls on Election Day.
This is of particular importance because racial minorities are egregiously underrepresented
on election day. Increasing turnout of black voters during a Presidential election has important
implications for more accurately representing the demographic makeup of the United States.
10
Low-Income Respondents
A main reason cited by low income individuals that they did not vote was “work/too busy.”
Traditionally, low-income individuals are too focused on making ends meet to worry about who
will be the president in two months. This conundrum is well explained by Theory of the Calculus
of Voting which states that voters must balance civic engagement with their other priorities (Farber
2009). Correspondingly, “work/too busy” was met with the most positive response among low-
income registered voters. We observe a 50% increase in potential voter turnout among low-income
individuals who cite work or too busy as the primary reason they were unable to vote.
High-Income Respondents
Again drawing on the census data we observe that the primary reason high-income individuals
failed to vote was “work” (U.S. Census Bureau 2010). We believe this is because of the work ethic
among high-income voters. This voting bloc typically prioritizes work over voting, so even though
they are probably able to incur the cost of voting, the benefit of voting does not adequately
compensate for the high cost of taking time off of work (Ibid.). We predict that eliminating this
10
We believe it is important to note that congress would correspondingly become more representative as ballots do
not only ask for voter’s preferred presidential candidate. This is not the focus of the paper, however, we suspect that
overall representation (at all levels of government up for election) would become much more demographically
representative.
Bradfield Johnson 20
cost (as many people would get the day off and not have to work) would potentially increase
turnout a full 55% among high-income voters who fail to vote due to “work/too busy.”
Moderate Respondents
Literature consistently claims that moderates are the least likely group to vote, which leads to
polarized candidates and polarized political leaders. Additionally, it is intuitive that moderates also
claim a higher cost to vote than ideologues because moderates tend to be less invested in politics.
We expect that removing the costs of voting will improve turnout for moderates as we
found it would for Black voters. We find impressive results with moderates who claimed “not
enough time/long lines” as the primary reason they failed to vote achieving a potential increase of
38%. Those who cited “work/too busy” would potentially see turnout increase by up to 52%.
Of particular interest are moderates who cited “forgetting to vote.” Moderates are probably
the most likely group to forget about Election Day. Making Election Day a national holiday would
make it very difficult to forget. Indeed, we could see an increase of up to 45% among this group
of moderates.
Rural Respondents
The final subgroup of interest are those respondents that live in rural areas (as compared to
suburban and urban areas). We hypothesize that those who live in rural areas have higher cost to
voting because of the distance to polling locations. Thus we believe that the largest impact would
be on the group who cited “not enough time/long lines.” We found that turnout among rural voters
would increase by up to 49%. If voters in rural areas did not have other demands and long drives
to polling places might seem a little more worth it.
Bradfield Johnson 21
Placebo Tests
For our placebo test, we ran the second logit model again and included a variable that, according
to our theoretical framework, would not be correlated to making Election Day a national holiday.
If this variable was statistically insignificant, then we could be sure respondents were not just
claiming making Election Day a national holiday would be helpful independent of reasons they
failed to vote. We found the test variable to be statistically insignificant with a p-value of over 0.6.
Thus, our results have both statistical and substantive significance.
The first variable, “illness or disability” should not be affected by making Election Day a
national holiday. Changing the status of Election Day is unrelated to health, and thus the ill would
still fail to vote. Correspondingly we found not only were the results insignificant at the .05 level,
but were they significant there would be an inverse relationship with a coefficient of -0.134
(standard error of .324) meaning that making Election Day a National holiday would yield no
effect on the “ill or disabled” respondents.
Bradfield Johnson 22
Table 3: Placebo Test
Variable
Coef
Std Er
Illness or disability (First Response)
-.134
.324
Work/too busy (First Response)
.719*
.197
Family demands (First Response)
.744†
.370
Not enough time/long lines (First Response)
.655†
.324
Forgetting to vote (First Response)
.533‡
.315
Income
.098
.078
Rural/urban
.006
.107
Region
-.041
.059
Education
.051
.063
Ethnicity
.025
.079
Age
-.018
.007
Gender
-.199
.163
Ideology
-.167
.046
Constant
.821
.549
* is significant at the .01 level, †is significant at the .05 level and the ‡
is significant at the .1 level
Limitations
While the results that we found proved to be both statistically and substantively significant, the
study is not without its weaknesses. Our research is limited to a comparison of only one other
country: France. This poses several problems. First, the United States and France are not identical.
There may be some truths about French voters that cause them to turnout at higher rates than voters
in the United States for which we are unable to observe or control. However, these risks will always
be inherent in a case study. We minimize these risks by choosing a country that is most similar to
the United States. Further, we will take note of additional factors we discover that might result in
higher voter turnout.
Bradfield Johnson 23
Additionally, there may be problems with our survey. First, we are only asking three
questions. We combat this problem by being as specific as possible in the questions that we do
ask. There are also many options for respondents to choose from in the questions; we worry that
excessive options will distract respondents from accurately answering the question. However,
given our limits on how many questions we are allowed to ask, we believe these specific questions
are our best option.
While we do not have the resources necessary to mount a full scale experiment in which
we make Election Day a holiday in a small sample of United States precincts, our study yielded
limited, yet important, results. Internal validity is a limitation of particular concern. Since Election
Day has been on a Tuesday, we have no empirical evidence to which we can compare our results.
External validity is also a concern. Case studies have poor external validity, but we combat this by
using survey questions which will bridge the gap and make our results more externally valid.
Despite these limitations, we believe that our findings have implications for future
research. We suggest future research expand our findings through looking at all countries that have
made voting day either a weekend or a national holiday. Our findings, for the time being, can
create a stepping stone for future research and can work to persuade law-makers to consider the
potential impact that making Election Day a national holiday could have on increasing voter
turnout.
Conclusion
Abraham Lincoln once taught that a government “of the people, by the people, and for the people,
shall not perish from the Earth.” However, voting trends indicate that Americans are much less
involved in the political process than citizens of most developed democracies. If a government “by
Bradfield Johnson 24
the people” is so important to American political thought, then working to resolve the issue of
America’s dismal turnout is of utmost importance.
Many have suggested reasons as to why voting is important. MassVote, an organization
dedicated to improving voter turnout levels in the United States, has called voting “the cornerstone
of democracy” (MassVote 2016). Research suggests that that elected officials are most responsive
to likely voters (White, Nathan, and Faller 2014), thus implying that elected officials in a
democratic republic must have high turnout levels to accurately represent the will of the people.
From a moral perspective, one might argue that those with the ability to vote have a responsibility
to do so in order to exert their inherent right to influence government and policy.
Little work has been done to suggest ways to remedy the problem of low voter turnout in
America. Unlike other measures instituted to increase turnout, making Election Day a holiday
decrease the associated cost of voting and would be relatively easy to accomplish. Implementing
the simple creation of a national holiday for Election Day would improve turnout dramatically.
In our effort to determine the effect that a national holiday might have on voter turnout, we
conducted a case study and an original survey to provide statistical support to our hypothesis. We
found strong evidence suggesting that designating election day as a national holiday will have both
statistically and substantively significant results. Indeed, if America’s low voter turnout in the
United States is cause for concern, which we believe it must be, then the solution should be to
create a national holiday for Election Day. Doing so, we believe, will minimize many of the largest
barriers to voting.
Bradfield Johnson 25
Appendix
Table A1: Voting in France vs. Voting in the United States
France-Executive Elections 2002-2012
Year
Total Votes Cast
Registered Voters
Percent Registered Voters who voted
2012
37,016,309
46,066,307
80.35
2007
37,342,004
44,472,733
83.97
2002
32,832,295
41,191,169
79.71
107,190,608
131,730,209
81.37*
81.34**
United States-Executive Elections 2002-2012
Year
Total Votes Cast
Registered Voters
Percent Registered Voters who voted
2012
129,085,403
193,653,908
66.66
2008
133,944,538
190,461,401
70.33
2004
125,736,000
142,070,000
88.5
388,765,941
526,185,309
73.88*
75.16**
*This number was achieved by adding the total votes cast in each election to the total number of voters registered in
each election. We considered using this method to control for population size from year to year but found the
difference between this measure and the other to be similar enough to use the other, as this is how it is more
commonly calculated in literature. **This is the average voter turnout over the three elections we observe in each
country. These are the numbers we use in our discussion.
Bradfield Johnson 26
Table A2:
Logit Model: Would making an Election Day a Holiday solve the problem?
Pseudo R2 = .223
Variable
Coef
Std Er
Work/too busy (First Response)
2.57*
.220
Family demands (First Response)
1.58*
.358
Not enough time/long lines (First Response)
2.25*
.355
Forgetting to vote (First Response)
1.67*
.314
Income
.058
.088
Rural/urban
.069
.120
Region
-.057
.066
Education
.068
.071
Ethnicity
.069
.089
Age
-.015
.008
Gender
.050
.182
Ideology
-.068
.051
Constant
-1.19
.617
* Significant at the .01 level. Controls included.
Bradfield Johnson 27
Figure A5: Reasons Respondents or Respondent’s Neighbors Failed to vote
We found there was no statistical difference between our variables of significance so we combined the
responses. Only “other” was found to be statistically different between the two groups.
Bradfield Johnson 28
References
Ansolabehere, Stephen, and Brian Schaffner. "CCES Common Content, 2012 - CCES
Dataverse". Dataverse.Harvard.Edu. Last modified 2016. Accessed February 4, 2016.
https://dataverse.harvard.edu/dataset.xhtml?persistentId=hdl:1902.1/21447.
Ansolabehere, Stephen, and Charles Stewart III. "Waiting In Line To Vote". Support The Voter.
Last modified 2013. Accessed February 4, 2016.
https://www.supportthevoter.gov/files/2013/08/Waiting-in-Line-to-Vote-White-Paper-
Stewart-Ansolabehere.pdf.
Conley, Richard S. "Presidential Republics And Divided Government: Lawmaking And
Executive Politics In The United States And France". Political Science Quarterly 122, no. 2
(2007): 257-285.
Country-facts.findthedata.com,. "United States Vs France - Country Facts Comparison". Last
modified 2016. Accessed February 18, 2016. http://country-
facts.findthedata.com/compare/1-98/United-States-vs-France.
Downs, Anthony. An Economic Theory of Democracy, New York, Harper & Brothers, 1957.
Farber, Henrey. "Rational Choice And Voter Turnout: Evidence From Union Representation
Elections" (2016).
Gans, Curtis. "2012 Election Turnout Dips Below 2008 And 2004 Levels: Number Of Eligible
Voters Increases By Eight Million, Five Million Fewer Votes Cast". Bipartisan Policy
Center. Last modified 2016. Accessed February 4, 2016. http://bipartisanpolicy.org/wp-
content/uploads/sites/default/files/2012%20Voter%20Turnout%20Full%20Report.pdf.
Gimpel, James G., Joshua J. Dyck, and Daron R. Shaw. "Location, Knowledge And Time
Pressures In The Spatial Structure Of Convenience Voting". Electoral Studies 25, no. 1
(2006): 35-58.
Gomez, Brad T., Thomas G. Hansford, and George A. Krause. "The Republicans Should Pray
For Rain: Weather, Turnout, And Voting In U.S. Presidential Elections". J of Pol 69, no. 03
(2007).
Haspel, Moshe, and H. Gibbs Knotts. "Location, Location, Location: Precinct Placement And
The Cost Of Voting". Cambridge University Press. Last modified 2005. Accessed February
4, 2016. http://www.appstate.edu/~ehrhardtgc/Haspel--Location.pdf.
Heppner, P. Paul, Bruce E Wampold, and Dennis M Kivlighan. Research Design In Counseling.
Belmont, CA: Thomson Brooks/Cole, 2008.
Hirczy, Wolfgang. "The Impact Of Mandatory Voting Laws On Turnout: A Quasi-Experimental
Approach". Electoral Studies 13, no. 1 (1994): 64-76.
Bradfield Johnson 29
Holtz‐Bacha, Christina, Lynda Lee Kaid, and Anne Johnston. "Political Television Advertising
In Western Democracies: A Comparison Of Campaign Broadcasts In The United States,
Germany, And France". Political Communication 11, no. 1 (1994): 67-80.
Huffstutter, P.J. "Officials Face Election Day Stumper, With Possible Payoff Online". The
Seattle Times. Last modified 2006. Accessed April 8, 2016.
http://www.seattletimes.com/nation-world/officials-face-election-day-stumper-with-possible-
payoff-online/.
Idea.int,. "Voter Turnout Data For Germany (Parliamentary, EU Parliament) | Voter Turnout |
International IDEA". Last modified 2016. Accessed February 4, 2016.
http://www.idea.int/vt/countryview.cfm?id=61.
Jackman, Robert W. "Political Institutions And Voter Turnout In The Industrial Democracies".
The American Political Science Review 81, no. 2 (1987): 405.
King, James D. "Political Culture, Registration Laws, And Voter Turnout Among The American
States". Publius 24, no. 4 (1994): 115.
Linternaute.com,. "Bureau De Vote : Connartre Adresse Et Horaires". Last modified 2016.
Accessed February 4, 2016. http://www.linternaute.com/ville/p-bureau-vote#trouver-bureau-
vote.
Keller, Michael, and Yvette Romero. "The Definitely Messy, Probably Solvable Reasons
Americans Don't Vote". Bloomberg.Com. Last modified 2016. Accessed April 7, 2016.
http://www.bloomberg.com/politics/graphics/2016-non-voters/.
Nationale, Assemblace. "Voting - General Elections In France - 2012". Elections-Legislatives.Fr.
Last modified 2016. Accessed February 4, 2016. http://www.elections-
legislatives.fr/en/voting.asp.
Old Farmer's Almanac,. "Weather History By State". Last modified 2016. Accessed February 4,
2016. http://www.almanac.com/weather/history/states.
Pierce, Roy. Choosing The Chief. Ann Arbor: University of Michigan Press, 1995.
Piven, Frances Fox, Richard A Cloward, and Frances Fox Piven. Why Americans Still Don't
Vote. Boston: Beacon Press, 2000.
Powell, G. Bingham. "American Voter Turnout In Comparative Perspective". The American
Political Science Review 80, no. 1 (1986): 17.
Rayman, Ann, Rob O'Dell, and Ricardo Cano. "Arizona Primary: Maricopa County Had One
Polling Site For Every 21,000 Voters". Azcentral. Last modified 2016. Accessed April 7,
2016. http://www.azcentral.com/story/news/politics/elections/2016/03/22/live-arizona-
primary-coverage-presidential-preference-election/82096726/.
Stein, Robert M., and Greg Vonnahme. "Engaging The Unengaged Voter: Vote Centers And
Voter Turnout". jop 70, no. 02 (2008).
Bradfield Johnson 30
Washington Post. Last modified 2016. Accessed February 4, 2016.
https://www.washingtonpost.com/news/the-fix/wp/2015/07/17/why-dont-americans-vote-
were-too-busy/.
Wattenberg, Martin P. "Should Election Day Be A Holiday? - 98.10". Theatlantic.Com. Last
modified 2016. Accessed February 4, 2016.
http://www.theatlantic.com/past/issues/98oct/electday.htm.
Weatheronline.co.uk,. "France History - Graph - Weatheronline". Last modified 2016. Accessed
February 4, 2016.
http://www.weatheronline.co.uk/weather/maps/forecastmaps?CONT=euro&LAND=FR&R=
150.
White, Ariel R., Noah L. Nathan, and Julie K. Faller. "What Do I Need To Vote? Bureaucratic Discretion
And Discrimination By Local Election Officials". Am Polit Sci Rev 109, no. 01 (2014): 129-142.
