SlideShare a Scribd company logo
1 of 19
Download to read offline
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 0
All Vote-by-Mail Elections and
Voter Turnout in Utah
MPP Capstone Project
3/23/2016
University of Utah
Trent Matheson
Image by Rick Bowmer at the Standard Examiner
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 1
All Vote-by-Mail Elections and Voter Turnout in Utah:
Executive Summary
Background:
 In 2012, the Utah State Legislature gave county election administrators the authority to choose to
conduct their elections in an all vote-by-mail (all-VBM) format in place of the more traditional multi-
method elections.
 In the 2014 general election, 10 counties conducted all-VBM election and 19 counties plan to conduct
all-VBM election for the 2016 general election.
 Since the mid-1990s Utah has been trending below the national average for voter turnout.
 All-VBM format elections are often seen by policymakers as a remedy to low voter turnout.
 Evidence from research of all-VBM elections having an effect on voter turnout in Oregon, Washington,
Colorado, and California has been mixed.
 Past research has suggested that there are novelty effects on voter turnout where it only improved
voter turnout temporarily.
Focus of this Research:
 This paper is seeking to quantify the effects on voter turnout in even-year general elections between
2000- 2014 as a result of counties changing election formats.
 A linear regression analysis of all-VBM format elections on voter turnout was conducted while
controlling for effects from party competitiveness, presidential elections, and proportion of the
population that lives in an urban area.
Key Finding:
 Counties that conducted all-VBM format federal elections had 4.6% higher registered voter turnout
than the counties that had the traditional mix of polling place and absentee format elections.
Conclusions and Recommendations:
 Countywide all-VBM format elections are still new enough to Utah that there are likely novelty
effects on voter turnout.
 If increasing voter turnout is the objective of all-VBM elections then Utah should stay the course of
allowing counties to choose their election format and continue to study it.
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 2
Table of Contents:
Executive Summary......................................................................................................................... 1
Introduction:................................................................................................................................... 3
All-VBM in Other States:................................................................................................................. 4
All-VBM in Utah: ............................................................................................................................. 5
Theoretical Framework:.................................................................................................................. 8
Previous Research:.......................................................................................................................... 9
Data and Methods: ....................................................................................................................... 11
Results:.......................................................................................................................................... 14
Conclusion:.................................................................................................................................... 15
References: ................................................................................................................................... 17
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 3
Introduction: In the 1980s Utah’s eligible voter turnout was well above the national average,
ranking as high as 5th
in the nation. Beginning in 1996, Utah consistently ranked below the
national average in both presidential and midterm elections. By 2014 Utah was ranking near
the bottom (McDonald, 2016). This trend is illustrated in Figure 1 and is especially apparent in
midterm election turnout. With voter turnout on the decline in recent years, Utah policymakers
have been looking for policy options to improve voter turnout. Conducting elections in an all
vote-by-mail (all-VBM) format in place of more traditional election formats has been promoted
as a possible solution. In the 2012 legislative session, counties were given the option to choose
to conduct their elections in an all-VBM format. The impact of this policy change on voter
turnout is what this paper will be examining.
Figure 1: Data from the United States Election Project at the University of Florida
All-VBM elections have garnered a lot of press in recent years touting a positive effect on voter
turnout. This has especially been the case with improved voter turnout numbers in counties
conducting all-VBM elections in the 2015 municipal general elections. Was this due to the
institution of an all-VBM policy or due to other factors? Although the focus of this paper will be
on federal general elections, this analysis will still be seeking to quantify the effect size of all-
VBM elections and separating it from other factors that also correlate with voter turnout. When
voter turnout percentages are discussed, the absolute number will always be used and not the
year over year change. For example: if registered voter turnout increased in Davis County from
15% to 25%, we would say a 10% increase occurred and not a 60% increase. The year over year
measurement makes it difficult to keep track of the proportion of the change.
0%
10%
20%
30%
40%
50%
60%
70%
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014
General Election Voter Turnout of Eligible Voting
Population 1980-2014
US Presidential Utah Presidential US Midterm Utah Midterm
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 4
All-VBM advocates argue that the change in election format increases voter turnout by making
it easier for voters to cast their ballots as well as providing a reminder to voters that an election
is occurring. However, the empirical evidence for these assertions is mixed (Kousser & Mullin,
2007; Gronke & Miller, 2012). The assumption is that the “costs” of voting by mail are less than
the “costs” of voting at a polling place and therefore voter turnout will increase. These costs are
not financial burdens, but rather the time and effort necessary to remember an election is
imminent and to ultimately cast a ballot. All-VBM elections also have the potential to reduce
the financial burden of elections. While this may be a convincing reason to switch to all-VBM
elections in and of itself, this paper will be focusing on voter turnout effects.
All-VBM in Other States:
Oregon, Washington, and Colorado have all made the change from traditional multi-method
elections to all-VBM elections statewide. At least nineteen other states allow for specific
elections or specific jurisdictions to conduct elections in an all-VBM format (Underhill, 2016).
Figure 2: Data from the National Conference of State Legislatures
Oregon was the first state to conduct an all-VBM election. In 1981 Oregon allowed specific local
elections to be conducted in an all-VBM format. In the 1990s the state conducted several
statewide all-VBM elections. By the year 2000, all elections were being conducted in an all-VBM
format (Gronke & Miller, 2012). Washington allowed counties to decide whether to conduct
elections in an all-VBM format beginning in 2005, and by 2012 it was implemented statewide
for all elections (Gerber, Huber, & Hill, 2013). Colorado began conducting all-VBM elections in
2013 as a result of a highly popular and successful no-excuse absentee voting program (Cann,
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 5
Hall, & Monson, 2014). No-excuse absentee voting, in comparison to other types of absentee
voting, allows any registered voter to request an absentee ballot and no excuse is required. This
type of absentee voting is allowed in Utah. Some states require a valid excuse to be presented
that would prevent you from voting at a polling location such as a disability, being elderly, or
having work obligations.
Colorado, Oregon, and Washington had very high absentee voter rates prior to switching to all-
VBM format elections. This made it relatively easy for election administrators to implement all-
VBM elections as it was often the next logical step and ultimately streamlined election
administration processes. A September 2014 study commissioned by Utah’s Lieutenant
Governor’s office found that Utah’s absentee voting rates are far lower than these states prior
to switching to all-VBM elections (Cann et al., 2014).
Colorado, Oregon, and Washington all enjoy higher than the national average voter turnout and
their residents currently turn out to vote more frequently than Utahns do. On the face of it,
policymakers may leap to the conclusion that this is due to the relatively unique election format
but this correlation is not necessarily causal. This paper seeks to systematically measure if there
is causality as well as the size and direction of the causality. For example, one reason that these
states may have higher voter turnout may be that the three
all-VBM states also currently have a more generally
competitive political environment than Utah or many other
states.
All-VBM in Utah: Prior to 2012, jurisdictions with very
small populations were able to decide whether they wanted
to conduct an election in an all-VBM format. Currently some
counties that do not have all-VBM format elections still have
some very rural all-VBM precincts. Beginning in the 2012
session with House Bill 172, the Utah State Legislature
began allowing counties to decide whether to hold elections
in an all-VBM format or in the more traditional format of a
combination of polling places, early voting, and no-excuse
absentee ballots (Bateman, 2015).
In 2012, Duchesne was the only county to conduct an all-
VBM format general election. For the 2014 general election,
ten counties conducted all-VBM format elections. In 2016,
Iron, Juab, Morgan, Rich, Salt Lake, Summit, Uintah,
Table 1: Counties Using All-
VBM in Federal General
Elections
2012 2014 2016
Duchesne Beaver
Cache
Davis
Duchesne
Garfield
Grand
San Juan
Sanpete
Sevier
Wayne
Beaver
Cache
Davis
Duchesne
Garfield
Grand
Iron
Juab
Morgan
Rich
Salt Lake
San Juan
Sanpete
Sevier
Summit
Uintah
Wasatch
Wayne
Weber
Total: 1 Total: 10 Total: 19
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 6
Wasatch, and Weber counties will join the ten counties from 2014. Daggett County was unsure
if they were going to conduct the 2016 general election in an all-VBM format at the time this
paper was written. The specific counties that use all-VBM elections can be seen in Table 1. This
information was gathered in my conversations with the Lt. Governor’s Office and county clerks
election offices to determine the years that all-VBM general elections were conducted.
The momentum seems to be behind switching to an all-VBM format with almost two-thirds of
counties planning to conduct them in 2016 and no counties have chosen to switch back to
previous voting methods. Figure 3 shows the proportion of Utah registered voters voting via
absentee ballots in general elections compared to the proportion of registered voters who live
in all-VBM counties and will likely receive mail ballots. This suggests that Utah may have a
massive increase in the number and proportion of absentee ballots in 2016 if they keep pace
with the current proportion of the population living in all-VBM counties.
Figure 3: Data from the Utah Lt. Governor’s Office
There has been some recent Utah specific all-VBM research. The 2014 Lieutenant Governor’s
Office report had a number of interesting findings. They focused their research on the local
rural jurisdictions that conducted all-VBM elections and they found some evidence of novelty
effects. Immediately after the change in election format voter turnout increased in the cities
1%
18%
58%
1% 0% 2% 1%
7% 10%
19%
35%
0%
10%
20%
30%
40%
50%
60%
70%
2000 2002 2004 2006 2008 2010 2012 2014 2016
Voters Casting Absentee Ballots in General Elections
Compared to Voters to Living in All-VBM Counties
Precentage of Registered Voters Living in an All-VBM County
Percentage of Registered Voters Casting Absentee Ballots
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 7
they studied by about 10% on average. After two or three all-VBM elections had been
conducted voter turnout trended to toward previous levels. However, the authors theorized
that the gains in municipal and special elections turnout may be more permanent (Cann et al.,
2014). The authors’ assertion is rooted in previous research where the relative salience of an
election plays a role in how much of an effect all-VBM format elections have on voter turnout.
High salience elections, such as presidential elections, do not see the same increase in voter
turnout from all-VBM format elections because voters are more engaged and likely to turn out
to the polls anyway. Low salience elections, such as special, primary, and municipal elections,
garner far less attention than a presidential election and a ballot arriving in the mail serves as a
reminder for some voters that an election is going to occur and thus may be less subject to
novelty effects (Karp & Banducci 2000; Kouser & Mullin, 2007; Southwell, 2009; Gronke &
Miller, 2012).
The authors also found a lack of demand from voters for all-VBM elections. In the three polls
they examined the vast majority were not in support of all-VBM elections. The questions were
worded slightly different for each poll but none of them found support for statewide all-VBM
elections above 20%. The most recent poll found that 3% of Utahns want switch to all-VBM
format elections for all elections while 87% want to keep the current mix of voting methods
(Cann et al., 2014).
Cann et al. (2014) also found that the county clerks offices who implement this policy have very
diverse opinions on administering all-VBM elections. A plurality of 42% the clerks contacted
said that costs to hold election an all-VBM election would increase. One third said it would
reduce costs. And one quarter said that costs would stay the same. When asked about the
relative ease of administering an all-VBM election, 50% of the clerks that responded said it
would be more difficult to administer. 29% said it would be the same to administer and 21%
said it would be easier to administer.
More recent research was completed by the Utah Foundation where they found upward trends
of voter turnout in municipal elections that were conducted in an all-VBM format for the first
time in 2015. This study did not attempt to differentiate the estimate of the effect of all-VBM
elections from other factors but the counties and cities that were studied had markedly higher
turnout in 2015 compared to 2013 (Bateman 2015). This would be consistent with previous
research though more municipal elections will need to be studied to see if voter turnout
regresses toward previous levels.
What we have learned from the report to the Lieutenant Governor and the report from the
Utah Foundation is that the evidence for all-VBM elections increasing voter turnout is tenuous,
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 8
election officials have mixed feelings about it, and Utahns are not necessarily demanding a
change in election format. That being said, a new round of research polling of election officials
and voters may be in order now since both groups have more experience with these types of
elections since that polling took place. Many counties are going to switch to all-VBM elections
for the 2016 general election so county clerks’ attitudes may have changed since that polling
was conducted. With those planned changes, more and more voters will likely be voting via
mail and may change their opinion on implementing it statewide. This may also close the gap in
mail ballot rates in Utah and the rates that Oregon, Washington, and Colorado voted at before
they switched to all-VBM elections statewide.
Theoretical Framework: Voter turnout is important topic of study as it is a keystone for
democratic representation. When more people turnout to vote policymakers have a better idea
what candidates and policies their constituencies favor. If voter turnout is low then there is
potential for these preferences to be skewed and over-represent one constituency over
another. Voter turnout also provides an indicator for legitimacy (or lack thereof) of the current
system of political institutions. A healthy democracy hinges on the political participation of
those being governed.
Much of the research on the topic of voter turnout as a whole is rooted in the assumptions of
rational choice theory. Proponents of all-VBM elections hypothesize that these types of
elections reduce the costs of voting on the electorate which would result in increased voter
turnout (Southwell, 2000). Downs, (1957) as well as Riker and Ordeshook (1968), are early
pioneers for the rational choice approach to the study of voter turnout. Though these authors
did not study all-VBM elections directly, their work and theories are a good jumping off point
for examining the literature and its theoretical framework.
Downs’ (1957) equation for voter turnout is 𝑅 = (𝐵 ∗ 𝑃) − 𝐶. Where R is the reward of
voting, B is the benefits reaped from your chosen candidate winning, P is the probability that
your vote is the deciding vote, and C is the cost of voting itself. As the likelihood that any
individual voter will cast the deciding vote in any given election is extraordinarily small, any
positive measurement for the cost variable leads to a negative reward measurement. The
implication of this is that Downs’ equation predicts that no one will vote. Obviously this theory
is problematic because voters continue so to show up to the polls.
Riker and Ordeshook (1968) try to remedy this voting paradox with their equation, 𝑅 = 𝐵 ∗
𝑃 − 𝐶 + 𝐷, where they add a D for a sense of civic duty. This paradox has not yet been
resolved in its entirety but scholars such as Mark Franklin (2004) have made strides in that
effort by extending Riker and Ordeshook’s work to include social context of voting. What is
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 9
notable here is that the cost variable in the decision to cast a vote is of prime importance and
thus has a strong presence throughout all-VBM election research.
Franklin’s (2004) work is important on three fronts as it applies to all-VBM elections. First, his
theoretical framework regarding the “character of elections” is important to demonstrate that
all-VBM elections are an electoral institutional structure that can have a cumulative effect on
voter turnout over time. Second, his work illustrates the methodological difficulties in
measuring a policy’s effect on voter turnout over time. Third, he provides insight into aggregate
vs individual levels of analysis, and their implications for causality. Though Franklin did not
study all-VBM elections specifically, his analysis of absentee voting across established
democracies did find a decrease in voter turnout when absentee voting was introduced. He
provided context to this finding by saying that policymakers generally implemented absentee
voting as a reaction to declining voter turnout. Despite this change in policy, voter turnout still
generally declined in countries that allowed absentee voting. Franklin did not characterize this
finding as causal, because if there was an effect at all his analysis may have not have been
sensitive enough to detect it. This is partially due to his data being aggregated country level
voter turnout data where he assigned dummy variables to the institutional factors and partially
due to the time-series analysis. Time-series analysis with aggregate data tends to have issues
with confounding variables and can make it difficult to tease out individual effects like absentee
voting effects on voter turnout. These types of confounding issues are found throughout the
vote by mail literature.
Previous Research: As is the case with much public policy research, it is impossible to test the
effect of all-VBM elections in a laboratory. Therefore the effects must be tested via quasi-
experimental methods. As a result, it can be difficult to delineate the dose and effect of all-VBM
elections due to the many voter turnout covariates and election specific contextual factors. This
is part of the reason that the literature has wildly varying types of results estimating all-VBM
elections on voter turnout, especially on even year federal elections. However, the evidence
does show that odd-year, local, primary, and special elections generally benefit from all-VBM
elections. Some research shows that this increase in all-VBM is a temporary novelty effect and
that voter turnout then regresses to previous levels.
Oregon and Washington were the earliest to move to all-VBM elections statewide and thus are
the most studied examples estimating voter turnout effects. Early research showed as much as
a 10% increase in voter turnout in Oregon as a result of conducting them in an all-VBM format
(Southwell & Burchett, 2000). However, later research has shown that the voter turnout
increase was a temporary three-election novelty period (Gronke & Miller, 2012). After the
novelty period, voter turnout then regressed to previous levels. As both of these states are in
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 10TablefromoriginallyfromGerber,Huber,Hill(2013)butexpandedwithadditionalsourcesandmynotes.Studieswithasterisks(*)arethestudieswiththemostconvincingevidence.
Gerber,
Huber,Hill
(2013)*
Gronkeand
Miller(2012)*
Laroccaand
Klemanski
(2011)
Bergmanand
Yates(2011)
Southwell
(2009)
Richey(2008)
Kousserand
Mullin(2007)*
Franklin
(2004)*
(Absentees)
Southwelland
Burchett
(2000)
Karpand
Banducci
(2000)*
Magleby
(1987)
StudyAuthors
Table2:
SurveyofAll-VBMLiterature
Washington;4
countiesfor
individual
sample
Oregon
US
Nationwide
5Northern
California
Counties
Oregon
Nationwide
California
Established
Democracies
Oregon
Oregon
California,
Oregon,
Washington
PlacesStudied
TS1996-
2010;
IM2006-10
1960-2010
2000,2004,
2008
2006and
2008
1980-2007
1982-2006
2000and
2002
1945-2000
1960-1996
1986-2000
1980-1984
YearsStudied
CountyTime
Seriesand
Individual
Matching
Statewide
TimeSeries
Nationwide
Individual
CrossSection
Individual
TimeSeries,
CrossSection
Statewide
TimeSeries
Statewide
TimeSeries,
CrossSection
Precinct
Matching
Countrywide
TimeSeries
Statewide
TimeSeries
PrecinctTime
Series
Statewide
TimeSeries,
CrossSection
Method
TS3%
3%-20%
3.90%
11%
-2.70%
1.80%
President
TS3%
0.70%
8.70%
-1.50%
-2.90%
Midterm
TS3%
-0.7%-7.5%
1.3%–14.5%
1.5%-26.5%
Other
Election
-13.20%
10.20%
19%
PooledAll
Elections
Intime-seriesfound3%increaseinall
elections.Onlymeasuredcountiesthat
changedafter2006forindividual
matching.Onlyfoundindividualeffectsfor
thosethatalreadyvotedregularly.
Extendedandcouldnotreplicate
SouthwellandBurchett’s(2000)findings.
Onlyturnoutincreasesinlocalandspecial
elections
CensusCPSdata,PresidentialElections.
Onlystatesinsamplewithall-VBMwere
ORandWA.
Lettingvotersknowaheadoftimethat
votingmethodwillchangehelpsmitigate
lowerturnout.
Walksbackher2000findingstosaythat
onlylocalandspecialelectionsare
affecteddramaticallyandthatother
electiontypesonlymodestincrease.
AuthorsaysUSturnoutaresultofresultof
onerousvotingprocedures.Oregononly
statewithallVBMinanalysis.
Nearest-neighbormatchingprecinctpairs.
Onlyincreasedturnoutinlocalandspecial
elections.Decreasedotherwise.
Foundevidenceofabsenteevoting
associatedwithdecreasingturnoutin
lowerhouseelectionsbecauseof
confoundingvariables.
Findingsclaimthatthereisa10%increase
forVBMelections.Onlystudiedfirst3
VBMelections.
Higherturnoutforlowstimuluselections
(local/primary).Doesnothelpthosenot
alreadypredisposedtovote.
Outlierstudy,VBMonlylegalforsome
localandspecial(ballotmeasures).
Comparingapplestoorangeswithout
havingVBMfederalelectionstocompare
to.
Notes
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 11
the Pacific Northwest and have generally competitive political environments, the results from
the change in election format may not necessarily be directly generalizable to the rest of the
country.
Table 2 provides an overview of some of the major works in all-VBM research. It shows how the
research changed over time as well as the size of the all-VBM effect on the elections the
authors examined. The studies with asterisks next to their name are the studies that provided
the most convincing evidence.
There are two things that are convincingly established in the literature as a whole. First, all-
VBM elections moderately increase turnout in low stimulus elections such as odd year
municipal elections and special elections. Second, is that all-VBM elections primarily help
people to turnout to vote that were already inclined to vote in the first place. This relationship
has mixed findings in the literature (Gerber et al., 2013, Karp and Banducci, 2000). Voter
turnout data does not provide evidence that turnout is any higher for non-voters in an all-VBM
general election (Gerber et al., 2013). There is also little evidence that all-VBM elections are a
policy that would actually significantly increase voter turnout regardless of election type. It may
be that voters have a standing decision of whether or not to vote much like Franklin (2004) had
found in his analysis. These two categories of findings in conjunction provide evidence that all-
VBM elections are not a panacea for low voter turnout. If policymakers want to increase voter
participation in low stimulus elections then that reason alone may justify a change in election
format. Though it is not a cure-all for voter turnout in all election types or for activating the
entire electorate like some all-VBM advocates make it out to be.
Data and Methods: This paper examines general elections at the county level for the years of
2000-2014. A county level analysis was selected instead of a precinct level analysis because the
data is more accessible as well as being more persistent over time. A state level analysis likely
would require that some statewide all-VBM elections be conducted in order to get reliable
results. In order to measure the size and direction of county level all-VBM election voter
turnout effects, the differences between counties and election types must be controlled for.
The county level of analysis plays a role in how the analysis was conducted which will be
apparent subsequent sections.
The Utah Lieutenant Governor’s Office maintains county level election results data. From the
results data the number of ballots returned in each county were used for the numerator in the
calculation for voter turnout. Also, the number of registered voters in each county provided by
this dataset was used for the denominator in the measure for voter turnout. This is
demonstrated in the equation below with B standing for the number of total ballots cast and E
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 12
standing for the number of registered voters or the size of the electorate. The result of this
equation was multiplied by 100 in the data to aid in interpreting effect sizes in regression
analysis. This causes the dependent variable coefficients to represent the variance as a
percentage of voter turnout.
𝑅𝑒𝑔𝑖𝑠𝑡𝑒𝑟𝑒𝑑 𝑉𝑜𝑡𝑒𝑟 𝑇𝑢𝑟𝑛𝑜𝑢𝑡 =
𝐵
𝐸
The measure of eligible voter turnout (VEP) is preferred over registered voter turnout but there
is not consistent Census data for all counties before 2010. To illustrate this, before the 2012
general election the Lieutenant Governor’s office chose to purge some 76,000 inactive voters
instead of removing voters systematically over time (Gerhke, 2012). Most of these voters were
from Salt Lake County. This has the potential to influence registered voter turnout
measurements upward and this is a weakness of the voter turnout variable that the author
recognizes.
The Lieutenant Governor’s office also provided a list of counties that conducted all-VBM
elections in 2012 and 2014. Using that list we created a dummy variable for the counties that
conducted all-VBM elections. This variable will serve as the primary independent variable. We
also created a dummy variable for the 2000, 2004, 2008, and 2012 presidential elections in the
dataset as these elections almost always have significantly higher turnout than mid-term
elections.
The measure of party competitiveness in each election was developed from the vote totals for
candidates in the election results datasets from the Lieutenant Governor’s Office. This measure
is derived from the total votes in a county for Republican congressional candidates and
Democratic congressional candidates. Congressional races were chosen because they occurred
in every election we are examining and they are either at, or near, the top of the ticket. As
these races often cross county lines, the variable measures the total number of votes that all
Republican or Democratic congressional candidates received in a given county. The equation
below shows how we calculated the party competitiveness variable.
𝑃𝑎𝑟𝑡𝑦 𝐶𝑜𝑚𝑝𝑒𝑡𝑖𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠 =
(𝑅 − 𝐷)
(𝑅 + 𝐷)
R is the total number of ballots Republican congressional candidates received in a county. D is
the total number of ballots Democratic congressional candidates received in a county. The
denominator is the sum of the total number of ballots for both Republican and Democratic
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 13
candidates. The numerator measures the total ballot difference between Republican and
Democratic candidates. The measurement is centered around 0 and with theoretical limits
being between 1 and -1. When the output of this formula is 0, the parties are evenly matched in
a county. When the measure is negative, Democratic candidates received more votes. When
the measure is positive, Republican candidates received more votes. This measure in our data
varied from between .85 in Paiute County in 2014 to -.61 in Carbon County in 2008.
The Gronke and Miller (2012) or the Southwell and Burchett (2000) measures of
competitiveness would have been preferable but the results were not statistically significant.
𝑉1
is the number of ballots for the winning candidate, 𝑉2
is the number of ballots for the losing
candidate, and B is the total number of ballots cast. Subtracting it from 1 scales the measure so
that the higher the number, the more competitive the election is. Their formula is intended to
be used on state level analyses which may be part of the reason that the results were not
statistically significant in this analysis. At the county level a losing top of the ticket candidate
can get the most votes in a county and still lose. This likely does not provide a reliable measure
of competitiveness because voters in a county are still likely aware of the competitive situation
of the state or district as a whole.
𝐺𝑟𝑜𝑛𝑘𝑒 & 𝑀𝑖𝑙𝑙𝑒𝑟 𝐶𝑜𝑚𝑝𝑒𝑡𝑖𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠 = 1 − (
𝑉1
− 𝑉2
𝐵
)
The urban and rural population Census datasets from 2000 and 2010 were used to estimate the
proportion the population in each county that lives in an urban setting. This variable is used to
control for the turnout and demographic differences between rural and urban counties. Rural
populations theoretically have the most to gain from an all-VBM format election because they
are more likely to be far away from a polling location. Because the Census only measures the
size of urban and rural populations during the decennial census, the years between 2000 and
2010 were interpolated and the years beyond 2010 were extrapolated. The theoretical limit of
this variable is between 0 and 1. The variable has measurements that range from 0 in counties
like Daggett, Garfield, Rich, and Wayne to .99 for counties like Salt Lake and Davis. This is
hopefully a good measure of the fixed effects in the counties as using dummy variables for the
counties individually created multicollinearity problems.
A linear regression analysis was performed on the dataset using registered voter turnout as the
dependent variable and all-VBM status, presidential election status, urban population
proportion, and party competitiveness as independent variables. The result of this analysis will
provide a measure of the size of the effect of all-VBM format elections on voter turnout in Utah
at the county level.
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 14
Results: The findings of the regression analysis are found in Table 3. The model allows this
analysis to reject the null hypothesis and shows that all-VBM does have an effect on voter
turnout. The all-VBM variable accounted for a 4.6% increase in registered voter turnout in the
counties that conducted their elections in that format in 2012 and 2014. This means that with
the county and election year differences being held constant, counties that chose to conduct an
all-VBM format general election had about a 4.6% higher voter turnout than the counties who
held elections primarily at a polling place. Statewide registered voter turnout in 2012 was
approximately 80% and in 2014 it was approximately 46%. In this analysis all counties that
conducted all-VBM format elections had registered voter turnout higher than the statewide
average.
Table 3:
Model of Registered Voter Turnout
Variable Name Coefficient Standard
Error
Significance Model Summary
All-VBM 4.56 2.68 .091 Adjusted 𝑅2
.554
Party
Competitiveness
3.90 1.7 .023
Standard Error
of the Estimate
8.41
Presidential Election 17.20 1.125
.000
Significance .000
Percentage Urban -10.93 1.604 .000
(Constant) 57.15 1.29 .000
The finding of this study is relatively close to the Gerber et al. (2013) Washington county level
findings where they found a 3% increase in registered voter turnout in counties that conducted
all-VBM elections for all election types. That being said, counties in Utah have only conducted
all-VBM format elections for a maximum of two even year general elections so it is possible that
novelty effects are occurring and voter turnout levels could regress toward previous turnout
levels as more all-VBM elections are conducted.
The p-value for the all-VBM variable was less than .1 measured at .091. This means that the
finding is within an acceptable range of statistical significance. If this type of analysis were to be
done again in the future after more all-VBM elections have occurred, then the p-value may
become more significant with a larger population of all-VBM elections to analyze.
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 15
Presidential elections accounted for approximately 17.2% average increase in voter turnout in
comparison to midterm federal elections. This is what would be expected and is graphically
illustrated in the Introduction section in Figure 1.
Party competitiveness accounted for about 4% of the variance in voter turnout. The model
shows that counties that voted for Republican congressional candidates at higher margins also
were more likely to turnout to vote. This may be due to voters preferring to vote for winning
candidates which often happen to be Republicans in the state of Utah. Alternatively, this could
be due to Democratic Party voters being less likely to vote because their candidates will likely
lose or Democratic voters’ susceptibility to policies that encourage voter suppression. More
research will be needed to confirm what factors are causing this effect.
The proportion of a county’s population that is urban accounted for about 11% of the variation
in voter turnout. With all other factors being equal, a county with 100% urban population
would have on average have 11 % lower voter turnout than a county with 0% urban population.
This variable is a good indicator for the demographic differences between urban rural counties
that scale with increasing urban populations such as income and fixed county effects that
influence voter turnout.
Table 4:
Measures of
Multicollinearity
Dependent Variables
Registered
Voter
Turnout
All-
VBM
Party
Competitiveness
Presidential
Election
Percentage
Urban
Independent
Variables
Registered Voter
Turnout
N/A 1.065 1.032 1.036 1.004
All-VBM 2.253 N/A 1.033 2.011 1.191
Party
Competitiveness
2.230 1.055 N/A 2.087 1.192
Presidential
Election
1.124 1.032 1.048 N/A 1.106
Percentage Urban 1.921 1.078 1.056 1.952 N/A
The model also showed no multicollinearity issues with the highest variance inflation factor
(VIF) measurement of 2.25. All measurements are well within an acceptable range and did not
have an effect on our coefficients. Table 4 shows the exact measurements.
Conclusion: This paper has demonstrated a 4.6% increase in registered voter turnout in
counties that conducted all-VBM elections. The all-VBM policy is still relatively new to Utah and
to ensure novelty effects are properly accounted for, researchers will need more elections to
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 16
study its effects. With the possibility of novelty effects being the source of the federal election
voter turnout increase, political efforts by the legislature may be better placed in other types of
election reform if increased voter turnout is the intended consequence. Furthermore,
additional yearly American Community Survey (ACS) Census datasets will likely be available in
the future so that all-VBM effects on voter turnout studies will be able to use voting eligible
voter turnout instead of, or in combination with, registered voter turnout.
All-VBM format elections can be a legitimate and valid way to conduct elections but the
decision to switch to all-VBM does not come without tradeoffs and costs. There is an entire
literature dedicated to what all-VBM format elections do and do not do well. There can be good
or bad polling place elections just like there can be good or bad all-VBM elections. The Navajo
Nation is suing San Juan County because all-VBM format elections put a disproportionately
severe burden on Navajo voters. San Juan County closed all but one of their polling locations
which forces Navajo voters to drive about 200 miles to the county seat if they want to vote in
person. Navajo voters often need help with their ballots because they have limited English
reading ability and Navajo is traditionally an unwritten language. Navajo voters assert that the
all-VBM system does not provide proper oral assistance to Navajo voters that have a limited
grasp of the English language (McFall, 2016). This type of issue is emblematic of the types of
tradeoffs that county election administrators have to take into account when they alter election
policies. Figure 4: Map of All-VBM Status in Utah
The uneven change to all-VBM format
elections can cause ethical problems for
election administrators with the principle of
one person, one vote. If cities choose to
conduct all-VBM format municipal elections
when there is a countywide seat or initiative,
those cities with all-VBM format elections
may have a disproportionate say in those
decisions. There are similar issues when
counties choose to conduct all-VBM format
elections for seats or initiatives that cross
county boundaries. It is possible that this will
serve as incentive for election administrators
to change election format.
All-VBM format elections alone will not
change the tide for declining voter turnout in
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 17
Utah. Additional voter engagement in elections, voter engagement in the candidate nominating
process, and more competitive elections are crucial pieces to solving the declining voter
turnout puzzle. Utah may be best suited to stay the course by allowing county clerks to decide
whether or not to hold elections in an all-VBM format or in the more traditional multi-method
elections. There is certainly momentum behind counties deciding to conduct all-VBM format
elections with at least 19 counties choosing to conduct them since the change in policy in 2012.
The position of county clerk in Utah is an elected office and voters are able to hold them
accountable for their decisions whether they do or do not like their election format. There may
be some promise that all-VBM elections will have a positive effect on Utah’s odd year municipal
elections but more study is needed. It may be wise for Utah to not move to all-VBM statewide
until there is good evidence that it increases voter turnout long term, a larger proportion of
voters choose to use mail ballots, or there is demand from voters to make the change.
References:
Bateman, M. (2015). Voting in Utah: Analyzing Current Practices and Future Options for Utah Voters.
Utah Foundation. Retrieved February 2, 2016, from
http://www.utahfoundation.org/reports/voting-in-utah-analyzing-current-practices-and-future-
options-for-utah-voters
Bergman, E., & Yates, P. A. (2011). Changing election methods: How does mandated vote-by-mail affect
individual registrants? Election Law Journal, 10(2), 115-127
Cann, D., Hall, T., & Monson, Q. (2014). Evaluating the Feasibility of Vote By Mail in Utah. Report for
Utah Lt. Governor's Office Elections Division. Retrieved February 3, 2016, from
http://elections.utah.gov/Media/Default/Leg Reports/VBM Study.pdf
Downs, A. (1957). An economic theory of democracy. New York, NY: Harper.
Franklin, M. N. (2004). Voter turnout and the dynamics of electoral competition in established
democracies since 1945. Cambridge University Press.
Gerber, A. S., Huber, G. A., & Hill, S. J. (2013). Identifying the effect of all-mail elections on turnout:
Staggered reform in the evergreen state. Political Science Research and Methods, 1(01), 91-116.
Gehrke, R. (2012, October 9). Utah purges tens of thousands from voter rolls. Salt Lake Tribune.
Retrieved March 3, 2016, from
http://archive.sltrib.com/story.php?ref=/sltrib/politics/55046504-90/county-democrats-
elections-officials.html.csp
Gronke, P., & Miller, P. (2012). Voting by Mail and Turnout in Oregon Revisiting Southwell and Burchett.
American Politics Research, 40(6), 976-997.
Karp, J. A., & Banducci, S. A. (2000). Going postal: How all-mail elections influence turnout. Political
Behavior, 22(3), 223-239.
Kousser, T., & Mullin, M. (2007). Does voting by mail increase participation? Using matching to analyze a
natural experiment. Political Analysis, 15(4), 428-445.
A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 18
Larocca, R., & Klemanski, J. S. (2011). US State Election Reform and Turnout in Presidential Elections.
State Politics & Policy Quarterly, 11(1), 76-101.
Magleby, D. (1987). Participation in mail ballot elections. Western Political Quarterly, 40(1), 79-91.
McDonald, M. (2016). Voter Turnout. United States Elections Project. Retrieved from March 8, 2016
from http://www.electproject.org/home/voter-turnout/voter-turnout-data.
McFall, M. (2016, February 26). San Juan County sued for violating Voting Rights Act. Salt Lake Tribune.
Retrieved April 11, 2016, from http://www.sltrib.com/news/3587049-155/san-juan-county-
sued-for-violating
Richey, S. (2008). Voting by mail: Turnout and institutional reform in Oregon. Social Science Quarterly,
89(4), 902-915.
Riker, W., & Ordeshook, P. (1968). A theory of the calculus of voting. American Political Science Review,
62(1), 25-42.
Southwell, P. (2009). Analysis of the turnout effects of vote by mail elections, 1980-2007. Social Science
Journal, 46(1), 211-217.
Southwell, P., & Burchett, J. (2000). The effect of all-mail elections on voter turnout. American Politics
Research, 28(1), 72-79.
Underhill, W. (2016, March 8). All-Mail Elections. National Conference of State Legislatures. Retrieved
March 9, 2016, from http://www.ncsl.org/research/elections-and-campaigns/all-mail-
elections.aspx

More Related Content

What's hot

Research Proposal on Voting Preferences
Research Proposal on Voting Preferences Research Proposal on Voting Preferences
Research Proposal on Voting Preferences Diana Dela Torre (Andi)
 
Referendum in Moldova: is it a failure and which are the implications?
Referendum in Moldova: is it a failure and which are the implications?Referendum in Moldova: is it a failure and which are the implications?
Referendum in Moldova: is it a failure and which are the implications?IDIS Viitorul
 
AS Government & Politics - Voting and Social Influences
AS Government & Politics - Voting and Social InfluencesAS Government & Politics - Voting and Social Influences
AS Government & Politics - Voting and Social Influenceskirstyodair
 
National Civic Summit - Lorraine C. Minnite
National Civic Summit - Lorraine C. MinniteNational Civic Summit - Lorraine C. Minnite
National Civic Summit - Lorraine C. MinniteNational Civic Summit
 
Brazilian Politics: Between Chaos and Stagnation
Brazilian Politics: Between Chaos and StagnationBrazilian Politics: Between Chaos and Stagnation
Brazilian Politics: Between Chaos and Stagnationfhguarnieri
 
Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...
Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...
Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...The Journal of Social Sciences Research
 
The Electoral Process
The Electoral ProcessThe Electoral Process
The Electoral Processkbeacom
 
Crime stats pressreleasecius09
Crime stats pressreleasecius09Crime stats pressreleasecius09
Crime stats pressreleasecius09sevans-idaho
 
Helping Wisconsin Vote
Helping Wisconsin VoteHelping Wisconsin Vote
Helping Wisconsin Votelfschmidli
 
The quality of government and why elections fail ( pdf drive )
The quality of government and why elections fail ( pdf drive )The quality of government and why elections fail ( pdf drive )
The quality of government and why elections fail ( pdf drive )Shobhit Thapliyal
 
Regular elections and multi party democracy elly twi may 2014
Regular elections and multi party democracy elly twi may 2014Regular elections and multi party democracy elly twi may 2014
Regular elections and multi party democracy elly twi may 2014Elly Twineyo Kamugisha
 
AndersonFall2012_geog591
AndersonFall2012_geog591AndersonFall2012_geog591
AndersonFall2012_geog591Lauren Anderson
 
American Prohibition Secondary Analysis & Unit Of Analysis[1]
American Prohibition Secondary Analysis & Unit Of Analysis[1]American Prohibition Secondary Analysis & Unit Of Analysis[1]
American Prohibition Secondary Analysis & Unit Of Analysis[1]captureasmile
 
Big data platforms for a better politics - Paul Hilder (Crowdpac)
Big data platforms for a better politics - Paul Hilder (Crowdpac)Big data platforms for a better politics - Paul Hilder (Crowdpac)
Big data platforms for a better politics - Paul Hilder (Crowdpac)mysociety
 

What's hot (19)

Research Proposal on Voting Preferences
Research Proposal on Voting Preferences Research Proposal on Voting Preferences
Research Proposal on Voting Preferences
 
Referendum in Moldova: is it a failure and which are the implications?
Referendum in Moldova: is it a failure and which are the implications?Referendum in Moldova: is it a failure and which are the implications?
Referendum in Moldova: is it a failure and which are the implications?
 
AS Government & Politics - Voting and Social Influences
AS Government & Politics - Voting and Social InfluencesAS Government & Politics - Voting and Social Influences
AS Government & Politics - Voting and Social Influences
 
National Civic Summit - Lorraine C. Minnite
National Civic Summit - Lorraine C. MinniteNational Civic Summit - Lorraine C. Minnite
National Civic Summit - Lorraine C. Minnite
 
Counting the Council
Counting the CouncilCounting the Council
Counting the Council
 
Brazilian Politics: Between Chaos and Stagnation
Brazilian Politics: Between Chaos and StagnationBrazilian Politics: Between Chaos and Stagnation
Brazilian Politics: Between Chaos and Stagnation
 
Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...
Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...
Dealing with Ghana’s Winner-Takes-All Politics: A Case for Proportional Repre...
 
paper (Autosaved)
paper (Autosaved)paper (Autosaved)
paper (Autosaved)
 
The Electoral Process
The Electoral ProcessThe Electoral Process
The Electoral Process
 
Crime stats pressreleasecius09
Crime stats pressreleasecius09Crime stats pressreleasecius09
Crime stats pressreleasecius09
 
Helping Wisconsin Vote
Helping Wisconsin VoteHelping Wisconsin Vote
Helping Wisconsin Vote
 
Unit 2
Unit 2Unit 2
Unit 2
 
The quality of government and why elections fail ( pdf drive )
The quality of government and why elections fail ( pdf drive )The quality of government and why elections fail ( pdf drive )
The quality of government and why elections fail ( pdf drive )
 
Tania
TaniaTania
Tania
 
E314456.pdf
E314456.pdfE314456.pdf
E314456.pdf
 
Regular elections and multi party democracy elly twi may 2014
Regular elections and multi party democracy elly twi may 2014Regular elections and multi party democracy elly twi may 2014
Regular elections and multi party democracy elly twi may 2014
 
AndersonFall2012_geog591
AndersonFall2012_geog591AndersonFall2012_geog591
AndersonFall2012_geog591
 
American Prohibition Secondary Analysis & Unit Of Analysis[1]
American Prohibition Secondary Analysis & Unit Of Analysis[1]American Prohibition Secondary Analysis & Unit Of Analysis[1]
American Prohibition Secondary Analysis & Unit Of Analysis[1]
 
Big data platforms for a better politics - Paul Hilder (Crowdpac)
Big data platforms for a better politics - Paul Hilder (Crowdpac)Big data platforms for a better politics - Paul Hilder (Crowdpac)
Big data platforms for a better politics - Paul Hilder (Crowdpac)
 

Viewers also liked

Viewers also liked (17)

Evolución de la Web
Evolución de la WebEvolución de la Web
Evolución de la Web
 
Evolucion de la web
Evolucion de la webEvolucion de la web
Evolucion de la web
 
Semana 7 excel
Semana 7   excelSemana 7   excel
Semana 7 excel
 
Evolucion de la Web
Evolucion de la WebEvolucion de la Web
Evolucion de la Web
 
Web 1.0 a 5.0
Web 1.0 a 5.0Web 1.0 a 5.0
Web 1.0 a 5.0
 
Evolucion de la web
Evolucion de la webEvolucion de la web
Evolucion de la web
 
Evoluciondelawed fer
Evoluciondelawed ferEvoluciondelawed fer
Evoluciondelawed fer
 
Presentacion la evolucion de la web
Presentacion la evolucion de la webPresentacion la evolucion de la web
Presentacion la evolucion de la web
 
Evolucion de la web
Evolucion de la webEvolucion de la web
Evolucion de la web
 
Evoluciondelawed 140909124447-dianarosero
Evoluciondelawed 140909124447-dianaroseroEvoluciondelawed 140909124447-dianarosero
Evoluciondelawed 140909124447-dianarosero
 
Evolucion de la web
Evolucion de la webEvolucion de la web
Evolucion de la web
 
Evolucion de la web
Evolucion de la webEvolucion de la web
Evolucion de la web
 
Evolucion de la web
Evolucion de la webEvolucion de la web
Evolucion de la web
 
Evolucion de la Web
Evolucion de la WebEvolucion de la Web
Evolucion de la Web
 
Evolucion de la web
Evolucion de la web Evolucion de la web
Evolucion de la web
 
Evolucion de la web
Evolucion de la webEvolucion de la web
Evolucion de la web
 
Las palabras y las cosas
Las palabras y las cosasLas palabras y las cosas
Las palabras y las cosas
 

Similar to Matheson Capstone Final Draft V2

Predicting Voter Turnout in the 2008 Presidential Election
Predicting Voter Turnout in the 2008 Presidential ElectionPredicting Voter Turnout in the 2008 Presidential Election
Predicting Voter Turnout in the 2008 Presidential Electionjemccull
 
Chapter 7 presentation
Chapter 7 presentationChapter 7 presentation
Chapter 7 presentationkrobinette
 
Report electoral institutions and political competition
Report electoral institutions and political competitionReport electoral institutions and political competition
Report electoral institutions and political competitionArtemSakh
 
Down-home Democracy: Why covering local elections is important and how small ...
Down-home Democracy: Why covering local elections is important and how small ...Down-home Democracy: Why covering local elections is important and how small ...
Down-home Democracy: Why covering local elections is important and how small ...Reynolds Journalism Institute (RJI)
 
Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...
Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...
Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...Toni Menninger
 
The haves and the have nots: is civic tech impacting the people who need it m...
The haves and the have nots: is civic tech impacting the people who need it m...The haves and the have nots: is civic tech impacting the people who need it m...
The haves and the have nots: is civic tech impacting the people who need it m...mysociety
 
The Main Characteristics Of Alternative Vote, PR List...
The Main Characteristics Of  Alternative Vote, PR List...The Main Characteristics Of  Alternative Vote, PR List...
The Main Characteristics Of Alternative Vote, PR List...Sonia Sanchez
 
Capstone Thesis Paper
Capstone Thesis PaperCapstone Thesis Paper
Capstone Thesis PaperEmily Greene
 
Vote.org 2016 Impact Report
Vote.org 2016 Impact ReportVote.org 2016 Impact Report
Vote.org 2016 Impact ReportVote.org
 
Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612
Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612
Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612Magellan Strategies
 
Keeping College Grads in VT
Keeping College Grads in VTKeeping College Grads in VT
Keeping College Grads in VTLaura Felone
 
The Case for Mobile Voting
The Case for Mobile VotingThe Case for Mobile Voting
The Case for Mobile VotingWagner College
 
Dan Centinello: Has the U.S Improved the Way It Runs Elections?
Dan Centinello: Has the U.S Improved the Way It Runs Elections?Dan Centinello: Has the U.S Improved the Way It Runs Elections?
Dan Centinello: Has the U.S Improved the Way It Runs Elections?Dan Centinello
 
A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...
A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...
A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...SIR SUCCESS PRINCE DUAH DUAH
 

Similar to Matheson Capstone Final Draft V2 (20)

Predicting Voter Turnout in the 2008 Presidential Election
Predicting Voter Turnout in the 2008 Presidential ElectionPredicting Voter Turnout in the 2008 Presidential Election
Predicting Voter Turnout in the 2008 Presidential Election
 
Chapter 7 presentation
Chapter 7 presentationChapter 7 presentation
Chapter 7 presentation
 
Report electoral institutions and political competition
Report electoral institutions and political competitionReport electoral institutions and political competition
Report electoral institutions and political competition
 
Down-home Democracy: Why covering local elections is important and how small ...
Down-home Democracy: Why covering local elections is important and how small ...Down-home Democracy: Why covering local elections is important and how small ...
Down-home Democracy: Why covering local elections is important and how small ...
 
Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...
Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...
Descriptive and Inferential Statistical Methods: Analysis of Voting and Elect...
 
Complete Study
Complete StudyComplete Study
Complete Study
 
The haves and the have nots: is civic tech impacting the people who need it m...
The haves and the have nots: is civic tech impacting the people who need it m...The haves and the have nots: is civic tech impacting the people who need it m...
The haves and the have nots: is civic tech impacting the people who need it m...
 
The Main Characteristics Of Alternative Vote, PR List...
The Main Characteristics Of  Alternative Vote, PR List...The Main Characteristics Of  Alternative Vote, PR List...
The Main Characteristics Of Alternative Vote, PR List...
 
Capstone Thesis Paper
Capstone Thesis PaperCapstone Thesis Paper
Capstone Thesis Paper
 
Vote.org 2016 Impact Report
Vote.org 2016 Impact ReportVote.org 2016 Impact Report
Vote.org 2016 Impact Report
 
Andre blais 1
Andre blais 1Andre blais 1
Andre blais 1
 
Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612
Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612
Magellan Strategies 2012 Internal Survey Research Summary Memorandum 120612
 
Keeping College Grads in VT
Keeping College Grads in VTKeeping College Grads in VT
Keeping College Grads in VT
 
Scott Swafford, Introduction, overview and research
Scott Swafford, Introduction, overview and researchScott Swafford, Introduction, overview and research
Scott Swafford, Introduction, overview and research
 
The Case for Mobile Voting
The Case for Mobile VotingThe Case for Mobile Voting
The Case for Mobile Voting
 
Dan Centinello: Has the U.S Improved the Way It Runs Elections?
Dan Centinello: Has the U.S Improved the Way It Runs Elections?Dan Centinello: Has the U.S Improved the Way It Runs Elections?
Dan Centinello: Has the U.S Improved the Way It Runs Elections?
 
Cameroon--2011 Global FDA Electoral Fairness Audit Report
Cameroon--2011 Global FDA Electoral Fairness Audit ReportCameroon--2011 Global FDA Electoral Fairness Audit Report
Cameroon--2011 Global FDA Electoral Fairness Audit Report
 
Door Knocker Essay
Door Knocker EssayDoor Knocker Essay
Door Knocker Essay
 
A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...
A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...
A CRITICAL STUDY OF THE IMPACT OF UNEMPLOYMENT ON VOTER TURNOUT IN SOME SELEC...
 
panthers07
panthers07panthers07
panthers07
 

Matheson Capstone Final Draft V2

  • 1. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 0 All Vote-by-Mail Elections and Voter Turnout in Utah MPP Capstone Project 3/23/2016 University of Utah Trent Matheson Image by Rick Bowmer at the Standard Examiner
  • 2. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 1 All Vote-by-Mail Elections and Voter Turnout in Utah: Executive Summary Background:  In 2012, the Utah State Legislature gave county election administrators the authority to choose to conduct their elections in an all vote-by-mail (all-VBM) format in place of the more traditional multi- method elections.  In the 2014 general election, 10 counties conducted all-VBM election and 19 counties plan to conduct all-VBM election for the 2016 general election.  Since the mid-1990s Utah has been trending below the national average for voter turnout.  All-VBM format elections are often seen by policymakers as a remedy to low voter turnout.  Evidence from research of all-VBM elections having an effect on voter turnout in Oregon, Washington, Colorado, and California has been mixed.  Past research has suggested that there are novelty effects on voter turnout where it only improved voter turnout temporarily. Focus of this Research:  This paper is seeking to quantify the effects on voter turnout in even-year general elections between 2000- 2014 as a result of counties changing election formats.  A linear regression analysis of all-VBM format elections on voter turnout was conducted while controlling for effects from party competitiveness, presidential elections, and proportion of the population that lives in an urban area. Key Finding:  Counties that conducted all-VBM format federal elections had 4.6% higher registered voter turnout than the counties that had the traditional mix of polling place and absentee format elections. Conclusions and Recommendations:  Countywide all-VBM format elections are still new enough to Utah that there are likely novelty effects on voter turnout.  If increasing voter turnout is the objective of all-VBM elections then Utah should stay the course of allowing counties to choose their election format and continue to study it.
  • 3. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 2 Table of Contents: Executive Summary......................................................................................................................... 1 Introduction:................................................................................................................................... 3 All-VBM in Other States:................................................................................................................. 4 All-VBM in Utah: ............................................................................................................................. 5 Theoretical Framework:.................................................................................................................. 8 Previous Research:.......................................................................................................................... 9 Data and Methods: ....................................................................................................................... 11 Results:.......................................................................................................................................... 14 Conclusion:.................................................................................................................................... 15 References: ................................................................................................................................... 17
  • 4. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 3 Introduction: In the 1980s Utah’s eligible voter turnout was well above the national average, ranking as high as 5th in the nation. Beginning in 1996, Utah consistently ranked below the national average in both presidential and midterm elections. By 2014 Utah was ranking near the bottom (McDonald, 2016). This trend is illustrated in Figure 1 and is especially apparent in midterm election turnout. With voter turnout on the decline in recent years, Utah policymakers have been looking for policy options to improve voter turnout. Conducting elections in an all vote-by-mail (all-VBM) format in place of more traditional election formats has been promoted as a possible solution. In the 2012 legislative session, counties were given the option to choose to conduct their elections in an all-VBM format. The impact of this policy change on voter turnout is what this paper will be examining. Figure 1: Data from the United States Election Project at the University of Florida All-VBM elections have garnered a lot of press in recent years touting a positive effect on voter turnout. This has especially been the case with improved voter turnout numbers in counties conducting all-VBM elections in the 2015 municipal general elections. Was this due to the institution of an all-VBM policy or due to other factors? Although the focus of this paper will be on federal general elections, this analysis will still be seeking to quantify the effect size of all- VBM elections and separating it from other factors that also correlate with voter turnout. When voter turnout percentages are discussed, the absolute number will always be used and not the year over year change. For example: if registered voter turnout increased in Davis County from 15% to 25%, we would say a 10% increase occurred and not a 60% increase. The year over year measurement makes it difficult to keep track of the proportion of the change. 0% 10% 20% 30% 40% 50% 60% 70% 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 General Election Voter Turnout of Eligible Voting Population 1980-2014 US Presidential Utah Presidential US Midterm Utah Midterm
  • 5. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 4 All-VBM advocates argue that the change in election format increases voter turnout by making it easier for voters to cast their ballots as well as providing a reminder to voters that an election is occurring. However, the empirical evidence for these assertions is mixed (Kousser & Mullin, 2007; Gronke & Miller, 2012). The assumption is that the “costs” of voting by mail are less than the “costs” of voting at a polling place and therefore voter turnout will increase. These costs are not financial burdens, but rather the time and effort necessary to remember an election is imminent and to ultimately cast a ballot. All-VBM elections also have the potential to reduce the financial burden of elections. While this may be a convincing reason to switch to all-VBM elections in and of itself, this paper will be focusing on voter turnout effects. All-VBM in Other States: Oregon, Washington, and Colorado have all made the change from traditional multi-method elections to all-VBM elections statewide. At least nineteen other states allow for specific elections or specific jurisdictions to conduct elections in an all-VBM format (Underhill, 2016). Figure 2: Data from the National Conference of State Legislatures Oregon was the first state to conduct an all-VBM election. In 1981 Oregon allowed specific local elections to be conducted in an all-VBM format. In the 1990s the state conducted several statewide all-VBM elections. By the year 2000, all elections were being conducted in an all-VBM format (Gronke & Miller, 2012). Washington allowed counties to decide whether to conduct elections in an all-VBM format beginning in 2005, and by 2012 it was implemented statewide for all elections (Gerber, Huber, & Hill, 2013). Colorado began conducting all-VBM elections in 2013 as a result of a highly popular and successful no-excuse absentee voting program (Cann,
  • 6. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 5 Hall, & Monson, 2014). No-excuse absentee voting, in comparison to other types of absentee voting, allows any registered voter to request an absentee ballot and no excuse is required. This type of absentee voting is allowed in Utah. Some states require a valid excuse to be presented that would prevent you from voting at a polling location such as a disability, being elderly, or having work obligations. Colorado, Oregon, and Washington had very high absentee voter rates prior to switching to all- VBM format elections. This made it relatively easy for election administrators to implement all- VBM elections as it was often the next logical step and ultimately streamlined election administration processes. A September 2014 study commissioned by Utah’s Lieutenant Governor’s office found that Utah’s absentee voting rates are far lower than these states prior to switching to all-VBM elections (Cann et al., 2014). Colorado, Oregon, and Washington all enjoy higher than the national average voter turnout and their residents currently turn out to vote more frequently than Utahns do. On the face of it, policymakers may leap to the conclusion that this is due to the relatively unique election format but this correlation is not necessarily causal. This paper seeks to systematically measure if there is causality as well as the size and direction of the causality. For example, one reason that these states may have higher voter turnout may be that the three all-VBM states also currently have a more generally competitive political environment than Utah or many other states. All-VBM in Utah: Prior to 2012, jurisdictions with very small populations were able to decide whether they wanted to conduct an election in an all-VBM format. Currently some counties that do not have all-VBM format elections still have some very rural all-VBM precincts. Beginning in the 2012 session with House Bill 172, the Utah State Legislature began allowing counties to decide whether to hold elections in an all-VBM format or in the more traditional format of a combination of polling places, early voting, and no-excuse absentee ballots (Bateman, 2015). In 2012, Duchesne was the only county to conduct an all- VBM format general election. For the 2014 general election, ten counties conducted all-VBM format elections. In 2016, Iron, Juab, Morgan, Rich, Salt Lake, Summit, Uintah, Table 1: Counties Using All- VBM in Federal General Elections 2012 2014 2016 Duchesne Beaver Cache Davis Duchesne Garfield Grand San Juan Sanpete Sevier Wayne Beaver Cache Davis Duchesne Garfield Grand Iron Juab Morgan Rich Salt Lake San Juan Sanpete Sevier Summit Uintah Wasatch Wayne Weber Total: 1 Total: 10 Total: 19
  • 7. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 6 Wasatch, and Weber counties will join the ten counties from 2014. Daggett County was unsure if they were going to conduct the 2016 general election in an all-VBM format at the time this paper was written. The specific counties that use all-VBM elections can be seen in Table 1. This information was gathered in my conversations with the Lt. Governor’s Office and county clerks election offices to determine the years that all-VBM general elections were conducted. The momentum seems to be behind switching to an all-VBM format with almost two-thirds of counties planning to conduct them in 2016 and no counties have chosen to switch back to previous voting methods. Figure 3 shows the proportion of Utah registered voters voting via absentee ballots in general elections compared to the proportion of registered voters who live in all-VBM counties and will likely receive mail ballots. This suggests that Utah may have a massive increase in the number and proportion of absentee ballots in 2016 if they keep pace with the current proportion of the population living in all-VBM counties. Figure 3: Data from the Utah Lt. Governor’s Office There has been some recent Utah specific all-VBM research. The 2014 Lieutenant Governor’s Office report had a number of interesting findings. They focused their research on the local rural jurisdictions that conducted all-VBM elections and they found some evidence of novelty effects. Immediately after the change in election format voter turnout increased in the cities 1% 18% 58% 1% 0% 2% 1% 7% 10% 19% 35% 0% 10% 20% 30% 40% 50% 60% 70% 2000 2002 2004 2006 2008 2010 2012 2014 2016 Voters Casting Absentee Ballots in General Elections Compared to Voters to Living in All-VBM Counties Precentage of Registered Voters Living in an All-VBM County Percentage of Registered Voters Casting Absentee Ballots
  • 8. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 7 they studied by about 10% on average. After two or three all-VBM elections had been conducted voter turnout trended to toward previous levels. However, the authors theorized that the gains in municipal and special elections turnout may be more permanent (Cann et al., 2014). The authors’ assertion is rooted in previous research where the relative salience of an election plays a role in how much of an effect all-VBM format elections have on voter turnout. High salience elections, such as presidential elections, do not see the same increase in voter turnout from all-VBM format elections because voters are more engaged and likely to turn out to the polls anyway. Low salience elections, such as special, primary, and municipal elections, garner far less attention than a presidential election and a ballot arriving in the mail serves as a reminder for some voters that an election is going to occur and thus may be less subject to novelty effects (Karp & Banducci 2000; Kouser & Mullin, 2007; Southwell, 2009; Gronke & Miller, 2012). The authors also found a lack of demand from voters for all-VBM elections. In the three polls they examined the vast majority were not in support of all-VBM elections. The questions were worded slightly different for each poll but none of them found support for statewide all-VBM elections above 20%. The most recent poll found that 3% of Utahns want switch to all-VBM format elections for all elections while 87% want to keep the current mix of voting methods (Cann et al., 2014). Cann et al. (2014) also found that the county clerks offices who implement this policy have very diverse opinions on administering all-VBM elections. A plurality of 42% the clerks contacted said that costs to hold election an all-VBM election would increase. One third said it would reduce costs. And one quarter said that costs would stay the same. When asked about the relative ease of administering an all-VBM election, 50% of the clerks that responded said it would be more difficult to administer. 29% said it would be the same to administer and 21% said it would be easier to administer. More recent research was completed by the Utah Foundation where they found upward trends of voter turnout in municipal elections that were conducted in an all-VBM format for the first time in 2015. This study did not attempt to differentiate the estimate of the effect of all-VBM elections from other factors but the counties and cities that were studied had markedly higher turnout in 2015 compared to 2013 (Bateman 2015). This would be consistent with previous research though more municipal elections will need to be studied to see if voter turnout regresses toward previous levels. What we have learned from the report to the Lieutenant Governor and the report from the Utah Foundation is that the evidence for all-VBM elections increasing voter turnout is tenuous,
  • 9. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 8 election officials have mixed feelings about it, and Utahns are not necessarily demanding a change in election format. That being said, a new round of research polling of election officials and voters may be in order now since both groups have more experience with these types of elections since that polling took place. Many counties are going to switch to all-VBM elections for the 2016 general election so county clerks’ attitudes may have changed since that polling was conducted. With those planned changes, more and more voters will likely be voting via mail and may change their opinion on implementing it statewide. This may also close the gap in mail ballot rates in Utah and the rates that Oregon, Washington, and Colorado voted at before they switched to all-VBM elections statewide. Theoretical Framework: Voter turnout is important topic of study as it is a keystone for democratic representation. When more people turnout to vote policymakers have a better idea what candidates and policies their constituencies favor. If voter turnout is low then there is potential for these preferences to be skewed and over-represent one constituency over another. Voter turnout also provides an indicator for legitimacy (or lack thereof) of the current system of political institutions. A healthy democracy hinges on the political participation of those being governed. Much of the research on the topic of voter turnout as a whole is rooted in the assumptions of rational choice theory. Proponents of all-VBM elections hypothesize that these types of elections reduce the costs of voting on the electorate which would result in increased voter turnout (Southwell, 2000). Downs, (1957) as well as Riker and Ordeshook (1968), are early pioneers for the rational choice approach to the study of voter turnout. Though these authors did not study all-VBM elections directly, their work and theories are a good jumping off point for examining the literature and its theoretical framework. Downs’ (1957) equation for voter turnout is 𝑅 = (𝐵 ∗ 𝑃) − 𝐶. Where R is the reward of voting, B is the benefits reaped from your chosen candidate winning, P is the probability that your vote is the deciding vote, and C is the cost of voting itself. As the likelihood that any individual voter will cast the deciding vote in any given election is extraordinarily small, any positive measurement for the cost variable leads to a negative reward measurement. The implication of this is that Downs’ equation predicts that no one will vote. Obviously this theory is problematic because voters continue so to show up to the polls. Riker and Ordeshook (1968) try to remedy this voting paradox with their equation, 𝑅 = 𝐵 ∗ 𝑃 − 𝐶 + 𝐷, where they add a D for a sense of civic duty. This paradox has not yet been resolved in its entirety but scholars such as Mark Franklin (2004) have made strides in that effort by extending Riker and Ordeshook’s work to include social context of voting. What is
  • 10. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 9 notable here is that the cost variable in the decision to cast a vote is of prime importance and thus has a strong presence throughout all-VBM election research. Franklin’s (2004) work is important on three fronts as it applies to all-VBM elections. First, his theoretical framework regarding the “character of elections” is important to demonstrate that all-VBM elections are an electoral institutional structure that can have a cumulative effect on voter turnout over time. Second, his work illustrates the methodological difficulties in measuring a policy’s effect on voter turnout over time. Third, he provides insight into aggregate vs individual levels of analysis, and their implications for causality. Though Franklin did not study all-VBM elections specifically, his analysis of absentee voting across established democracies did find a decrease in voter turnout when absentee voting was introduced. He provided context to this finding by saying that policymakers generally implemented absentee voting as a reaction to declining voter turnout. Despite this change in policy, voter turnout still generally declined in countries that allowed absentee voting. Franklin did not characterize this finding as causal, because if there was an effect at all his analysis may have not have been sensitive enough to detect it. This is partially due to his data being aggregated country level voter turnout data where he assigned dummy variables to the institutional factors and partially due to the time-series analysis. Time-series analysis with aggregate data tends to have issues with confounding variables and can make it difficult to tease out individual effects like absentee voting effects on voter turnout. These types of confounding issues are found throughout the vote by mail literature. Previous Research: As is the case with much public policy research, it is impossible to test the effect of all-VBM elections in a laboratory. Therefore the effects must be tested via quasi- experimental methods. As a result, it can be difficult to delineate the dose and effect of all-VBM elections due to the many voter turnout covariates and election specific contextual factors. This is part of the reason that the literature has wildly varying types of results estimating all-VBM elections on voter turnout, especially on even year federal elections. However, the evidence does show that odd-year, local, primary, and special elections generally benefit from all-VBM elections. Some research shows that this increase in all-VBM is a temporary novelty effect and that voter turnout then regresses to previous levels. Oregon and Washington were the earliest to move to all-VBM elections statewide and thus are the most studied examples estimating voter turnout effects. Early research showed as much as a 10% increase in voter turnout in Oregon as a result of conducting them in an all-VBM format (Southwell & Burchett, 2000). However, later research has shown that the voter turnout increase was a temporary three-election novelty period (Gronke & Miller, 2012). After the novelty period, voter turnout then regressed to previous levels. As both of these states are in
  • 11. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 10TablefromoriginallyfromGerber,Huber,Hill(2013)butexpandedwithadditionalsourcesandmynotes.Studieswithasterisks(*)arethestudieswiththemostconvincingevidence. Gerber, Huber,Hill (2013)* Gronkeand Miller(2012)* Laroccaand Klemanski (2011) Bergmanand Yates(2011) Southwell (2009) Richey(2008) Kousserand Mullin(2007)* Franklin (2004)* (Absentees) Southwelland Burchett (2000) Karpand Banducci (2000)* Magleby (1987) StudyAuthors Table2: SurveyofAll-VBMLiterature Washington;4 countiesfor individual sample Oregon US Nationwide 5Northern California Counties Oregon Nationwide California Established Democracies Oregon Oregon California, Oregon, Washington PlacesStudied TS1996- 2010; IM2006-10 1960-2010 2000,2004, 2008 2006and 2008 1980-2007 1982-2006 2000and 2002 1945-2000 1960-1996 1986-2000 1980-1984 YearsStudied CountyTime Seriesand Individual Matching Statewide TimeSeries Nationwide Individual CrossSection Individual TimeSeries, CrossSection Statewide TimeSeries Statewide TimeSeries, CrossSection Precinct Matching Countrywide TimeSeries Statewide TimeSeries PrecinctTime Series Statewide TimeSeries, CrossSection Method TS3% 3%-20% 3.90% 11% -2.70% 1.80% President TS3% 0.70% 8.70% -1.50% -2.90% Midterm TS3% -0.7%-7.5% 1.3%–14.5% 1.5%-26.5% Other Election -13.20% 10.20% 19% PooledAll Elections Intime-seriesfound3%increaseinall elections.Onlymeasuredcountiesthat changedafter2006forindividual matching.Onlyfoundindividualeffectsfor thosethatalreadyvotedregularly. Extendedandcouldnotreplicate SouthwellandBurchett’s(2000)findings. Onlyturnoutincreasesinlocalandspecial elections CensusCPSdata,PresidentialElections. Onlystatesinsamplewithall-VBMwere ORandWA. Lettingvotersknowaheadoftimethat votingmethodwillchangehelpsmitigate lowerturnout. Walksbackher2000findingstosaythat onlylocalandspecialelectionsare affecteddramaticallyandthatother electiontypesonlymodestincrease. AuthorsaysUSturnoutaresultofresultof onerousvotingprocedures.Oregononly statewithallVBMinanalysis. Nearest-neighbormatchingprecinctpairs. Onlyincreasedturnoutinlocalandspecial elections.Decreasedotherwise. Foundevidenceofabsenteevoting associatedwithdecreasingturnoutin lowerhouseelectionsbecauseof confoundingvariables. Findingsclaimthatthereisa10%increase forVBMelections.Onlystudiedfirst3 VBMelections. Higherturnoutforlowstimuluselections (local/primary).Doesnothelpthosenot alreadypredisposedtovote. Outlierstudy,VBMonlylegalforsome localandspecial(ballotmeasures). Comparingapplestoorangeswithout havingVBMfederalelectionstocompare to. Notes
  • 12. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 11 the Pacific Northwest and have generally competitive political environments, the results from the change in election format may not necessarily be directly generalizable to the rest of the country. Table 2 provides an overview of some of the major works in all-VBM research. It shows how the research changed over time as well as the size of the all-VBM effect on the elections the authors examined. The studies with asterisks next to their name are the studies that provided the most convincing evidence. There are two things that are convincingly established in the literature as a whole. First, all- VBM elections moderately increase turnout in low stimulus elections such as odd year municipal elections and special elections. Second, is that all-VBM elections primarily help people to turnout to vote that were already inclined to vote in the first place. This relationship has mixed findings in the literature (Gerber et al., 2013, Karp and Banducci, 2000). Voter turnout data does not provide evidence that turnout is any higher for non-voters in an all-VBM general election (Gerber et al., 2013). There is also little evidence that all-VBM elections are a policy that would actually significantly increase voter turnout regardless of election type. It may be that voters have a standing decision of whether or not to vote much like Franklin (2004) had found in his analysis. These two categories of findings in conjunction provide evidence that all- VBM elections are not a panacea for low voter turnout. If policymakers want to increase voter participation in low stimulus elections then that reason alone may justify a change in election format. Though it is not a cure-all for voter turnout in all election types or for activating the entire electorate like some all-VBM advocates make it out to be. Data and Methods: This paper examines general elections at the county level for the years of 2000-2014. A county level analysis was selected instead of a precinct level analysis because the data is more accessible as well as being more persistent over time. A state level analysis likely would require that some statewide all-VBM elections be conducted in order to get reliable results. In order to measure the size and direction of county level all-VBM election voter turnout effects, the differences between counties and election types must be controlled for. The county level of analysis plays a role in how the analysis was conducted which will be apparent subsequent sections. The Utah Lieutenant Governor’s Office maintains county level election results data. From the results data the number of ballots returned in each county were used for the numerator in the calculation for voter turnout. Also, the number of registered voters in each county provided by this dataset was used for the denominator in the measure for voter turnout. This is demonstrated in the equation below with B standing for the number of total ballots cast and E
  • 13. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 12 standing for the number of registered voters or the size of the electorate. The result of this equation was multiplied by 100 in the data to aid in interpreting effect sizes in regression analysis. This causes the dependent variable coefficients to represent the variance as a percentage of voter turnout. 𝑅𝑒𝑔𝑖𝑠𝑡𝑒𝑟𝑒𝑑 𝑉𝑜𝑡𝑒𝑟 𝑇𝑢𝑟𝑛𝑜𝑢𝑡 = 𝐵 𝐸 The measure of eligible voter turnout (VEP) is preferred over registered voter turnout but there is not consistent Census data for all counties before 2010. To illustrate this, before the 2012 general election the Lieutenant Governor’s office chose to purge some 76,000 inactive voters instead of removing voters systematically over time (Gerhke, 2012). Most of these voters were from Salt Lake County. This has the potential to influence registered voter turnout measurements upward and this is a weakness of the voter turnout variable that the author recognizes. The Lieutenant Governor’s office also provided a list of counties that conducted all-VBM elections in 2012 and 2014. Using that list we created a dummy variable for the counties that conducted all-VBM elections. This variable will serve as the primary independent variable. We also created a dummy variable for the 2000, 2004, 2008, and 2012 presidential elections in the dataset as these elections almost always have significantly higher turnout than mid-term elections. The measure of party competitiveness in each election was developed from the vote totals for candidates in the election results datasets from the Lieutenant Governor’s Office. This measure is derived from the total votes in a county for Republican congressional candidates and Democratic congressional candidates. Congressional races were chosen because they occurred in every election we are examining and they are either at, or near, the top of the ticket. As these races often cross county lines, the variable measures the total number of votes that all Republican or Democratic congressional candidates received in a given county. The equation below shows how we calculated the party competitiveness variable. 𝑃𝑎𝑟𝑡𝑦 𝐶𝑜𝑚𝑝𝑒𝑡𝑖𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠 = (𝑅 − 𝐷) (𝑅 + 𝐷) R is the total number of ballots Republican congressional candidates received in a county. D is the total number of ballots Democratic congressional candidates received in a county. The denominator is the sum of the total number of ballots for both Republican and Democratic
  • 14. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 13 candidates. The numerator measures the total ballot difference between Republican and Democratic candidates. The measurement is centered around 0 and with theoretical limits being between 1 and -1. When the output of this formula is 0, the parties are evenly matched in a county. When the measure is negative, Democratic candidates received more votes. When the measure is positive, Republican candidates received more votes. This measure in our data varied from between .85 in Paiute County in 2014 to -.61 in Carbon County in 2008. The Gronke and Miller (2012) or the Southwell and Burchett (2000) measures of competitiveness would have been preferable but the results were not statistically significant. 𝑉1 is the number of ballots for the winning candidate, 𝑉2 is the number of ballots for the losing candidate, and B is the total number of ballots cast. Subtracting it from 1 scales the measure so that the higher the number, the more competitive the election is. Their formula is intended to be used on state level analyses which may be part of the reason that the results were not statistically significant in this analysis. At the county level a losing top of the ticket candidate can get the most votes in a county and still lose. This likely does not provide a reliable measure of competitiveness because voters in a county are still likely aware of the competitive situation of the state or district as a whole. 𝐺𝑟𝑜𝑛𝑘𝑒 & 𝑀𝑖𝑙𝑙𝑒𝑟 𝐶𝑜𝑚𝑝𝑒𝑡𝑖𝑡𝑖𝑣𝑒𝑛𝑒𝑠𝑠 = 1 − ( 𝑉1 − 𝑉2 𝐵 ) The urban and rural population Census datasets from 2000 and 2010 were used to estimate the proportion the population in each county that lives in an urban setting. This variable is used to control for the turnout and demographic differences between rural and urban counties. Rural populations theoretically have the most to gain from an all-VBM format election because they are more likely to be far away from a polling location. Because the Census only measures the size of urban and rural populations during the decennial census, the years between 2000 and 2010 were interpolated and the years beyond 2010 were extrapolated. The theoretical limit of this variable is between 0 and 1. The variable has measurements that range from 0 in counties like Daggett, Garfield, Rich, and Wayne to .99 for counties like Salt Lake and Davis. This is hopefully a good measure of the fixed effects in the counties as using dummy variables for the counties individually created multicollinearity problems. A linear regression analysis was performed on the dataset using registered voter turnout as the dependent variable and all-VBM status, presidential election status, urban population proportion, and party competitiveness as independent variables. The result of this analysis will provide a measure of the size of the effect of all-VBM format elections on voter turnout in Utah at the county level.
  • 15. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 14 Results: The findings of the regression analysis are found in Table 3. The model allows this analysis to reject the null hypothesis and shows that all-VBM does have an effect on voter turnout. The all-VBM variable accounted for a 4.6% increase in registered voter turnout in the counties that conducted their elections in that format in 2012 and 2014. This means that with the county and election year differences being held constant, counties that chose to conduct an all-VBM format general election had about a 4.6% higher voter turnout than the counties who held elections primarily at a polling place. Statewide registered voter turnout in 2012 was approximately 80% and in 2014 it was approximately 46%. In this analysis all counties that conducted all-VBM format elections had registered voter turnout higher than the statewide average. Table 3: Model of Registered Voter Turnout Variable Name Coefficient Standard Error Significance Model Summary All-VBM 4.56 2.68 .091 Adjusted 𝑅2 .554 Party Competitiveness 3.90 1.7 .023 Standard Error of the Estimate 8.41 Presidential Election 17.20 1.125 .000 Significance .000 Percentage Urban -10.93 1.604 .000 (Constant) 57.15 1.29 .000 The finding of this study is relatively close to the Gerber et al. (2013) Washington county level findings where they found a 3% increase in registered voter turnout in counties that conducted all-VBM elections for all election types. That being said, counties in Utah have only conducted all-VBM format elections for a maximum of two even year general elections so it is possible that novelty effects are occurring and voter turnout levels could regress toward previous turnout levels as more all-VBM elections are conducted. The p-value for the all-VBM variable was less than .1 measured at .091. This means that the finding is within an acceptable range of statistical significance. If this type of analysis were to be done again in the future after more all-VBM elections have occurred, then the p-value may become more significant with a larger population of all-VBM elections to analyze.
  • 16. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 15 Presidential elections accounted for approximately 17.2% average increase in voter turnout in comparison to midterm federal elections. This is what would be expected and is graphically illustrated in the Introduction section in Figure 1. Party competitiveness accounted for about 4% of the variance in voter turnout. The model shows that counties that voted for Republican congressional candidates at higher margins also were more likely to turnout to vote. This may be due to voters preferring to vote for winning candidates which often happen to be Republicans in the state of Utah. Alternatively, this could be due to Democratic Party voters being less likely to vote because their candidates will likely lose or Democratic voters’ susceptibility to policies that encourage voter suppression. More research will be needed to confirm what factors are causing this effect. The proportion of a county’s population that is urban accounted for about 11% of the variation in voter turnout. With all other factors being equal, a county with 100% urban population would have on average have 11 % lower voter turnout than a county with 0% urban population. This variable is a good indicator for the demographic differences between urban rural counties that scale with increasing urban populations such as income and fixed county effects that influence voter turnout. Table 4: Measures of Multicollinearity Dependent Variables Registered Voter Turnout All- VBM Party Competitiveness Presidential Election Percentage Urban Independent Variables Registered Voter Turnout N/A 1.065 1.032 1.036 1.004 All-VBM 2.253 N/A 1.033 2.011 1.191 Party Competitiveness 2.230 1.055 N/A 2.087 1.192 Presidential Election 1.124 1.032 1.048 N/A 1.106 Percentage Urban 1.921 1.078 1.056 1.952 N/A The model also showed no multicollinearity issues with the highest variance inflation factor (VIF) measurement of 2.25. All measurements are well within an acceptable range and did not have an effect on our coefficients. Table 4 shows the exact measurements. Conclusion: This paper has demonstrated a 4.6% increase in registered voter turnout in counties that conducted all-VBM elections. The all-VBM policy is still relatively new to Utah and to ensure novelty effects are properly accounted for, researchers will need more elections to
  • 17. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 16 study its effects. With the possibility of novelty effects being the source of the federal election voter turnout increase, political efforts by the legislature may be better placed in other types of election reform if increased voter turnout is the intended consequence. Furthermore, additional yearly American Community Survey (ACS) Census datasets will likely be available in the future so that all-VBM effects on voter turnout studies will be able to use voting eligible voter turnout instead of, or in combination with, registered voter turnout. All-VBM format elections can be a legitimate and valid way to conduct elections but the decision to switch to all-VBM does not come without tradeoffs and costs. There is an entire literature dedicated to what all-VBM format elections do and do not do well. There can be good or bad polling place elections just like there can be good or bad all-VBM elections. The Navajo Nation is suing San Juan County because all-VBM format elections put a disproportionately severe burden on Navajo voters. San Juan County closed all but one of their polling locations which forces Navajo voters to drive about 200 miles to the county seat if they want to vote in person. Navajo voters often need help with their ballots because they have limited English reading ability and Navajo is traditionally an unwritten language. Navajo voters assert that the all-VBM system does not provide proper oral assistance to Navajo voters that have a limited grasp of the English language (McFall, 2016). This type of issue is emblematic of the types of tradeoffs that county election administrators have to take into account when they alter election policies. Figure 4: Map of All-VBM Status in Utah The uneven change to all-VBM format elections can cause ethical problems for election administrators with the principle of one person, one vote. If cities choose to conduct all-VBM format municipal elections when there is a countywide seat or initiative, those cities with all-VBM format elections may have a disproportionate say in those decisions. There are similar issues when counties choose to conduct all-VBM format elections for seats or initiatives that cross county boundaries. It is possible that this will serve as incentive for election administrators to change election format. All-VBM format elections alone will not change the tide for declining voter turnout in
  • 18. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 17 Utah. Additional voter engagement in elections, voter engagement in the candidate nominating process, and more competitive elections are crucial pieces to solving the declining voter turnout puzzle. Utah may be best suited to stay the course by allowing county clerks to decide whether or not to hold elections in an all-VBM format or in the more traditional multi-method elections. There is certainly momentum behind counties deciding to conduct all-VBM format elections with at least 19 counties choosing to conduct them since the change in policy in 2012. The position of county clerk in Utah is an elected office and voters are able to hold them accountable for their decisions whether they do or do not like their election format. There may be some promise that all-VBM elections will have a positive effect on Utah’s odd year municipal elections but more study is needed. It may be wise for Utah to not move to all-VBM statewide until there is good evidence that it increases voter turnout long term, a larger proportion of voters choose to use mail ballots, or there is demand from voters to make the change. References: Bateman, M. (2015). Voting in Utah: Analyzing Current Practices and Future Options for Utah Voters. Utah Foundation. Retrieved February 2, 2016, from http://www.utahfoundation.org/reports/voting-in-utah-analyzing-current-practices-and-future- options-for-utah-voters Bergman, E., & Yates, P. A. (2011). Changing election methods: How does mandated vote-by-mail affect individual registrants? Election Law Journal, 10(2), 115-127 Cann, D., Hall, T., & Monson, Q. (2014). Evaluating the Feasibility of Vote By Mail in Utah. Report for Utah Lt. Governor's Office Elections Division. Retrieved February 3, 2016, from http://elections.utah.gov/Media/Default/Leg Reports/VBM Study.pdf Downs, A. (1957). An economic theory of democracy. New York, NY: Harper. Franklin, M. N. (2004). Voter turnout and the dynamics of electoral competition in established democracies since 1945. Cambridge University Press. Gerber, A. S., Huber, G. A., & Hill, S. J. (2013). Identifying the effect of all-mail elections on turnout: Staggered reform in the evergreen state. Political Science Research and Methods, 1(01), 91-116. Gehrke, R. (2012, October 9). Utah purges tens of thousands from voter rolls. Salt Lake Tribune. Retrieved March 3, 2016, from http://archive.sltrib.com/story.php?ref=/sltrib/politics/55046504-90/county-democrats- elections-officials.html.csp Gronke, P., & Miller, P. (2012). Voting by Mail and Turnout in Oregon Revisiting Southwell and Burchett. American Politics Research, 40(6), 976-997. Karp, J. A., & Banducci, S. A. (2000). Going postal: How all-mail elections influence turnout. Political Behavior, 22(3), 223-239. Kousser, T., & Mullin, M. (2007). Does voting by mail increase participation? Using matching to analyze a natural experiment. Political Analysis, 15(4), 428-445.
  • 19. A l l V o t e - b y - M a i l E l e c t i o n s a n d V o t e r T u r n o u t i n U t a h | 18 Larocca, R., & Klemanski, J. S. (2011). US State Election Reform and Turnout in Presidential Elections. State Politics & Policy Quarterly, 11(1), 76-101. Magleby, D. (1987). Participation in mail ballot elections. Western Political Quarterly, 40(1), 79-91. McDonald, M. (2016). Voter Turnout. United States Elections Project. Retrieved from March 8, 2016 from http://www.electproject.org/home/voter-turnout/voter-turnout-data. McFall, M. (2016, February 26). San Juan County sued for violating Voting Rights Act. Salt Lake Tribune. Retrieved April 11, 2016, from http://www.sltrib.com/news/3587049-155/san-juan-county- sued-for-violating Richey, S. (2008). Voting by mail: Turnout and institutional reform in Oregon. Social Science Quarterly, 89(4), 902-915. Riker, W., & Ordeshook, P. (1968). A theory of the calculus of voting. American Political Science Review, 62(1), 25-42. Southwell, P. (2009). Analysis of the turnout effects of vote by mail elections, 1980-2007. Social Science Journal, 46(1), 211-217. Southwell, P., & Burchett, J. (2000). The effect of all-mail elections on voter turnout. American Politics Research, 28(1), 72-79. Underhill, W. (2016, March 8). All-Mail Elections. National Conference of State Legislatures. Retrieved March 9, 2016, from http://www.ncsl.org/research/elections-and-campaigns/all-mail- elections.aspx