Local Officials as Partisan Operatives: An Examination of ... · Local Officials as Partisan Operatives: An Examination of Early Voting Sites in Texas Counties ... Markie McBrayer

Local Officials as Partisan Operatives: An Examinationof Early Voting Sites in Texas Counties

Andrea EckelmanDepartment of Political Science

University of Houston

Markie McBrayerDepartment of Political Science

University of Houston

R. Lucas WilliamsDepartment of Political Science

University of Houston

May 12, 2016

Abstract

We explore whether county commissioners follow their partisan allegiances when selecting voting sites. Be-

cause low-turnout elections are said to benefit Republicans and high-turnout elections to benefit Democrats,

we expect Republican-dominated commissions to create fewer early voting sites than Democrat-dominated

commissions. Similarly, because demographics are good predictors of citizen partisanship, we expect to

find Republican commissions select early voting sites that disenfranchise minority voters, while Democratic

commissions select voting locations more proximate to minority areas. Using an originally collected dataset,

we assess the party identification of county judges and commissioners and geo-code early voting locations

in Texas from the November 2014 election. First, we assess whether Republican commissions create fewer

sites in their counties than Democratic commissions. Second, we assess whether early voting locations are

equitably sited using spatial analyses in ArcMap. Third, we examine if the county commissioners court’s

partisan composition affects where the early voting sites are placed. Our results provide some support

for our theory that partisan county commissioners strategically create early voting sites to benefit their

party’s electoral fortunes; specifically, Republican commissions employ fewer early voting locations than

Democratic commissions. Yet, these locations do appear to be equitably sited. We discuss substantial and

statistical explanations for these results as well as their implications.

After the Supreme Court struck down the coverage formula in Section 4 of the Voting Rights Act

of 1965, several states, including Texas, moved quickly to pass new, more restrictive, voting laws.

Because parties vet candidates via the state nomination process, they can ensure party loyalists

occupy positions with key electoral powers, such as selecting early voting locations. In Texas,

the county commissioners court (composed of a county judge and four county commissioners)

has significant discretion in this regard, deciding where early voting locations will be placed. We

present three main reasons why county commissioners courts might strategically choose polling

sites: (1) they care about reelection and will utilize the tools of their office to enhance their chances

of winning, (2) they are progressively ambitious, and (3) parties select loyalists who will utilize

the powers of their office to bolster the party’s electoral prospects. We hypothesize that partisan

county commissioners will select locations that benefit their own party’s electoral prospects. To

that end, we seek to answer the question of how many sites the commissions will employ and

whether early voting sites are located in areas that disenfranchise minorities, given the close re-

lationship between race and partisanship in the electorate. There is some preliminary support for

our theoretical expectations. Specifically, county governments that are dominated by Republican

officials tend to have fewer early voting sites and less of their county’s population lives proximate

to such sites. Still, the early voting locations appear to be equitably sited in terms of race. Given

these results, we discuss the implications for democratic politics.

County Commission Description, Powers, and Responsibilities

Each state varies in how it selects early voting locations, and whether it allows early voting at all.1.

For instance, Washington, Oregon, and Colorado hold all elections by mail.2 In West Virginia, the

county commissioners select the election day sites, but early voting always occurs at the county

courthouse. North Carolina utilizes a State Board of Elections to appoint the county-level Board of

Elections, for which not more than two members may belong to the same party.3 Table 1 describes

how early voting is conducted in counties across the U.S.; whether county commissioners courts

1

(or the state’s equivalent) choose the location of in-person voting; and how many partisan county

commissioners are elected on a partisan ballot. In-Person Early Voting equals Y if counties in the

state conduct voting polls prior to election day. County prerogative equals Y if voting sites (in

general, not just early voting sites) are selected by county officials or the equivalent.4 Number (#)

of Partisan Counties equals the total count of counties in a state in which candidates for county-

wide office run in a primary and general election under partisan identities.5

In the State of Texas, each county commissioners court is made up of a county judge and four

county commissioners who share responsibility for placing early voting locations. Each of the

five positions are elected to four-year terms in partisan elections, and they comprise the governing

body of each county. County judges are elected by the county at-large and commissioners are

elected from four districts within each county. Party organizations in the state nominate members

to represent the party on the ballot. To be nominated, a prospective candidate must apply with

either a state- or county-level party chair. In terms of official responsibilities of commissioners

courts, county judges preside over the commissioners’ court and are the most powerful member.

Together, the judge and four county commissioners form the commissioners court, and they are

responsible for administering elections within the county. One of the more important functions of

the commissioners court is to determine where elections are held and how long the voting locations

are open.7 If the Texas county has a population less than 100,000, they need only have one early

voting location at the county election official’s main office, typically a city hall or court house,

meaning that in such a case the county government has no discretion in deciding its location.

Texas counties containing populations between 100,000 and 120,000 must set up at least one early

voting site in each county commissioner’s district if more than 15% of registered voters request

it.8 If the county’s population is between 120,000 and 400,000, then the county is required to have

the “main early voting polling place” to be operable for at least 12 hours each workday for the

final week of the early voting period, in addition to one site in each county commissioner’s district,

for a total of five early voting locations. Counties containing more than 400,000 residents have a

minimum of one early voting location per state legislative district, in addition to the main early

2

Table 1: Voting Location Determination by StateState In-Person Early Voting County Prerogative # Partisan Counties

Alabama N Y All (67)Alaska Y N 0 (10)

Arkansas Y Y All (75)California Y Y 0 (58)Colorado N N All (64)

Connecticut N - -Florida Y Y 62 (67)Georgia Y Y 157 (159)Idaho Y Y All (44)

Illinois Y Y All (102)Indiana Y Y All (92)

Iowa Y Y All (99)Kansas Y Y 103 (105)

Kentucky N Y All (120)Louisiana Y Y 0 (64)

Maine Y N All (16)Massachusetts N Y All (14)

Michigan N N All (83)Minnesota Y N 0 (87)Mississippi N Y All (82)

Montana Y N 46 (56)Nebraska Y Y All (93)Nevada Y Y All (17)

New Hampshire N N All (10)New Jersey Y Y All (21)

New Mexico Y Y All (33)New York N N All (57)

North Carolina Y N All (100)North Dakota Y Y 0 (53)

Oklahoma Y N All (77)Oregon N N 27 (36)

Pennsylvania N Y All (67)South Carolina N Y All (46)South Dakota Y Y 0 (66)

Tennessee Y N Some6

Texas Y Y All (254)Utah Y Y 27 (29)

Vermont Y N All (14)Virginia N Y All (95)

Washington N N 34 (39)West Virginia Y Y All (55)

Wisconsin Y N 0 (72)

voting location.9 There are 39 Texas counties with populations over 100,000, thus there are 39

county commissions with the potential discretion to decide early voting locations.

3

Party Organizations, Local Officials, and the Partisan Selection

of Polling Locations

We aim to determine whether county commissioners choose voting locations based on criteria

benefiting their own party in elections. Because local officials are beholden to their party organi-

zations, we argue that they toe the party line when the duties and powers of their office allow. For

instance, because county commissioners are elected on partisan ballots, they will be motivated to

look out for the interests of their party. Similarly to many state legislatures’ decennial redistricting

struggle, county commissions wish to shape the electoral landscape in a way that benefits their

own party. Commissioners’ desires to make it more convenient for their party’s “base” to vote

likely shape their decisions on where to place early voting locations and how many to place. We

expect these county officials to select voting sites that are geographically more proximate to these

demographic “bases.” We also expect Democratic county officials to place more sites overall in an

attempt to boost turnout.

The Costs of Voting, and Partisan Advantages in Turnout

According to Downs (1957), voters engage in a cost-benefit assessment of the decision to go to the

polls to vote. When perceived benefits outweigh the costs of casting a ballot, rational citizens vote.

This is, of course, a simplification of rational choice models of turnout. Aldrich (1993) explains

how the cost-benefit model is a practical tool for assessing the turnout phenomenon because simply

voting is one of the lowest-cost forms of participation. That is, casting a ballot is the most likely

form of political participation to experience higher benefits than costs. Among citizens choosing

not to turn out, the costs of voting that lead to “rational nonparticipation” include income, edu-

cation, institutional barriers, and the lack of resources associated with these factors (Rosenstone

and Hansen, 2003). In our examination of whether local party officials choose voting locations

that will benefit their own party, we simply assume increasing the costs of voting is a deterrent to

casting a ballot.

4

The vox populi regarding the partisan effects of turnout holds that large turnout benefits Demo-

cratic candidates and depressed turnout benefits Republicans. While research is mixed on the point

(e.g. Gomez, Hansford and Krause (2007); Hansford and Gomez (2010); Highton and Wolfinger

(2001); Tucker, Vedlitz and DeNardo (1986); DeNardo (1980)), we believe Republican and Demo-

cratic officeholders alike operate under the assumption that higher turnout benefits Democrats and

lower turnout benefits Republicans. Therefore, we need not assume this partisan turnout effect

to be the case; rather, we assume that local officeholders believe it to be the case. Laws thought

to depress turnout, such as voter ID restrictions, are often championed in Republican-dominated

state legislatures. Likewise, the National Voter Registration Act (NVRA) of 1993, passed by a

Democratically controlled Congress, reduced barriers to voting by requiring voter registration ca-

pabilities be made available at all benefit-administering agency offices. Due in part to the belief

that turnout benefits Democrats, Congress divided along party lines on this bill. Of the 73 original

co-sponsors of this bill, only one was Republican. Of the 62 senators voting Yea, only 5 were

Republican. Of the 259 House members voting Yea, only 21 were Republican. Not only does the

general public believe large turnout benefits Democrats while low turnout benefits Republicans,

but partisan officeholders at the highest level seem to believe it as well.

For locally elected partisan officials, increasing the costs of voting involves placing legal and

logistical barriers between the voter and the voting booth. Voter ID laws and registration deadlines

are examples of legal barriers to voting through which potential voters must navigate if they are

to cast a ballot on election day. Logistical barriers to voting include features of electoral admin-

istration, such as weekday voting, single-day voting, business-hours-only voting, and distance to

the polling place. Haspel and Knotts (2005) show that locating polling sites further away from

prospective voters reduces the likelihood of turnout. Other scholars find more nuance in this ar-

gument, noting a non-linear negative relationship between greater distances to in-person precinct

voting sites and low turnout. However, distance to a polling place is correlated with voting by mail

(Dyck and Gimpel, 2005). Dyck and Gimpel do find a negative linear relationship between turnout

and distance to early voting location. There are also empirical instances of local officials’ polling-

5

location decisions having implications for who turns out. For example, local officials moving

polling locations has disproportionately affected Democratic voters due to increased costs associ-

ated with the logistics of finding and securing transportation (Brady and McNulty, 2011). Other ev-

idence suggests that decreasing the costs of voting increases turnout (Stein and Vonnahme, 2008).

As we argue below, party operatives utilize the prerogatives of their office to construct logistical

barriers to voting (Republicans) and also to facilitate turnout (Democrats).

Local Officials as Party Operatives

In our explanation of why county commissions may select early voting sites that disenfranchise

minority citizens, we make several key assumptions. First, county commissioners are effectively

legislators. This is a widely accepted notion and is particularly accurate in Texas.10 Second, Downs

(1957) tells us that parties are rational end-maximizing agents. Parties are primarily made up of

election-oriented office seekers. We assume that, once in office, party members utilize available le-

gal tools and prerogatives to maximize the number of votes cast for their party on election day. One

of those tools, we argue, is the ability to select polling locations. Third, legislators are prospective

political actors concerned about their own political futures. Electoral concerns are therefore key to

a legislator’s decision-making calculus (Mayhew, 1974; Fenno, 1978). Politicians are strategic and

will behave in ways that benefit their party (Aldrich, 1993). In pursuit of the best possible election

circumstances for the future, county commissioners select early voting sites to benefit their party’s

electoral fortunes, in addition to their own.

If parties are rational end-maximizing agents (Downs, 1957), then they will utilize their or-

ganizational clout to motivate sub-state elected officials to make decisions strategically toward

the goal of benefiting their party’s electoral prospects. According to Aldrich (1995), contempo-

rary parties have become better parties-in-government and parties-as-organizations than parties-

in-the-electorate. As “parties in service to candidates,” political parties seek to boost the electoral

fortunes of office-seekers. In addition to services such as campaign bundling and training that

Aldrich (1995) mentions, one way parties can serve their candidates is through the ability to take

6

advantage of institutional prerogatives, such as voting site selection. By selecting polling sites

in locations more proximate to neighborhoods where residential demographics favor the party,

forward-thinking legislators can provide an additional edge for their party in the next election

cycle.

By providing resources, such as campaign financing and contacts with campaign managers,

parties become conduits for political careers. While the party organization operates in service of

popular candidates, ambitious candidates at the local level must prove themselves to the party.

“Nomination by a major party...remains an all but necessary condition for election to major of-

fice.” (Aldrich, 1995). Both progressively and statically ambitious politicians rely on party re-

sources to be elected and reelected. Politicians who are progressively ambitious plan to represent

the party in higher offices. Statically ambitious politicians wish to maintain their current positions

(Schlesinger, 1966).11 County commissioners who are progressively ambitious need to demon-

strate political skills and show service to the party if they are to run for higher office. One way

progressively ambitious commissioners might do so is by utilizing their office to boost electoral

fortunes for the party. In this way, commissioners serve the interests of their parties by strategically

locating voting sites to be most easily accessible to likely supporters of their party.

Party organizations also maintain close ties with their candidates. From recruitment to cam-

paign activities such as voter registration, local and state party organizations play a significant role

in the election of their candidates (Frendreis, Gibson and Vertz, 1990). Furthermore, legislative

leaders at the state level are often personally involved in recruiting candidates at the local level

(Sanbonmatsu, 2006). Party recruiters desire candidates who will support the party’s goals while

in office. Indeed, one survey of local party chairs in the South found they devote a large portion

of their efforts to activities, including candidate recruitment, which further the party’s interests

(Steed et al., 1998). This is particularly true for local party organizations and local political elites.

Masket (2011) describes informal party organizations (IPOs) as local elites connected to the state

and national party. These partisan gatekeepers control the nomination process. As noted above,

parties vet candidates via the nomination process, allowing the party to select loyalists to lower

7

level positions and more talented politicians to ambitious offices such as executive positions, con-

gressional offices, or the state legislature. When these potential loyalists hold local positions such

as county commissioner, they are expected to toe the party line when the responsibilities of their

office involve the electoral success of the party. For county commissioners in Texas, who are nom-

inated by state and local party officials, the ability to determine the quantity and location of early

voting sites is a useful tool in bolstering party electoral success.

Having established party organizations as a primary source of motivation for county commis-

sioners’ behavior and the partisan electoral consequences of turnout, we move toward developing

specific expectations about how legislators in such a party-dominated local political environment

might behave. It is well established that Republicans and Democrats have very different voting

bases in terms of racial and ethnic groups. 58% of non-Hispanic white Texans identify as Re-

publican and 28% as Democrat. Black Texans identify as Democrat at a rate of 76%, while only

9% are Republican. 25% of Hispanic Texans identify as Republican and 44% as Democrat. .12

Given these differences, parties do well to enable the socioeconomic demographics who generally

support them at the polls. We assert that placing early voting locations in select quantities and

nearer to neighborhoods containing favorable demographics is an electorally adroit maneuver that

one should expect of office seekers motivated by election-oriented party organizations.

Polling Site Quantity Hypothesis: If county commissioners behave as partisan operatives, then

Democratic counties will have more sites than Republican ones, since officials believe that lower-

ing the costs of voting benefits Democrats whereas increasing the costs of voting benefits Repub-

licans.

Polling Site Selection Hypothesis: If legislators are forward-looking, end-maximizing party op-

eratives, then Republican county commissioners will strategically locate early voting sites geo-

graphically proximate to white areas in order to bolster Republican candidate success. Likewise,

Democrats will locate early voting sites geographically proximate to non-white areas in order to

8

bolster Democratic candidate success.

Data

The locations of the early voting sites were obtained by searching county websites, as well as

media publications, for a list of locations and hours. For cases in which the site information was

obtained using media publications, we also contacted the county and confirmed the sites with them.

In total, we have sites from 39 Texas counties for a total of 740 full-time, temporary, and mobile

early voting locations. Our sample only includes counties with populations greater than 100,000

because only these counties have the discretion to site early voting locations. For counties with

populations less than 100,000, the early voting locations are required to be at the county election

official’s main office, typically a city hall or court house, meaning that the county government

has no discretion in deciding its location. There are three types of early voting locations in the

data, including full-time, temporary, and mobile early voting locations. From these we created two

types of early voting sites—those that are open all days or all but one or two, and those that are

open fewer days than that. For instance, in Travis County, some mobile voting locations are open

every single day of the early voting period, and, thus, we coded them as being full-time sites. For

those sites that were open less than that, we categorized them as temporary/mobile while noting

how many days they were open. In the first set of analyses that assess how partisanship affects

the number of early voting locations, we model the number of sites per county (broken out as all

sites, full-time, or temporary) as a function of the partisan composition of the county government.

We collected the partisan composition of the county government through Internet searches, as well

as by contacting the counties themselves. We ensured that the commissioners and judges were

in office by January 2014. This variable can take on values from 0 to 5—0 means that there

are only Democratic officials (commissioners and county judge) seated on the commission and 5

means that all commission officials are Republican. We expect a negative relationship between

county government and the number of sites, since Republican counties will attempt to increase the

9

costs of voting, thereby benefiting their party. We also control for the county’s population, on the

assumption that a greater county population is associated with more early voting sites.

In the second set of analyses, we assess how much of the county’s population lives within

one mile of an early voting location. In our collection of early voting locations, we also gathered

information on site, street address, city, and ZIP code. This information was then used to geo-

code the sites using Texas A&M’s publicly available batch geo-coding system. Approximately

four-fifths of the data were geo-coded correctly using parcel locations or street interpolation. The

other fifth were geo-coded by finding the ZIP code centroid, which did not meet the necessary

accuracy required for this project. For the cases in which the sites had obtained only the ZIP code

centroid, we turned to alternative means of geo-coding, specifically using Google’s latitude and

longitude statistics for these locations. This strategy ensures that all sites are geo-coded accurately.

Once the sites were obtained, we projected the locations in ArcMap onto Census block groups

with information on total population. The National Historical Geographic Information System

(NHGIS), managed by the University of Minnesota, provides an accessible interface for Census

data with geographic identifiers, and their site was used to obtain the block group shape file and

data. The block group shape file is from the 2010 TIGER Lines. In the analyses, we project 1-mile

buffers around each of the early voting sites. For an illustration of this strategy, see Figure 1, in

which early voting locations and their corresponding 1-mile buffer are mapped for central Houston

in Harris County.

Methodologically, we assume that individuals are distributed evenly across block groups. The

percent of the block group that lies within the buffer is multiplied by the total number of individu-

als. For instance, if 30% of one block group with a population of 100 and 40% of a second block

group with a population of 200 lies within the one-mile buffer, then the total population of the

buffer is 110 (.3*100+.4*200). We use this strategy to calculate the total number of individuals

within the buffer regions. For instance, if 60,000 people live within the 1-mile buffers of the early

voting locations, and the county is comprised of 400,000 individuals, then we can conclude that

15% of people have reasonable access to early voting locations. In the second analysis, we posit

10

Figure 1: Map of Houston Early Voting Sites with 1-Mile Buffer

11

that as a county commission becomes more conservative (or Republican), a smaller portion of the

population will live within the 1-mile buffer, meaning that more of the county’s population lacks

proximate access to early voting sites. We also control for the county’s total population, the density

of said population, and the area of the county. It is likely that counties with greater area and lower

density will see more of their population have proximate access to early voting sites.

In our third set of analyses, we assess where county governments will place their early voting

sites and which demographic communities have access to such sites. Again, we use the 1-mile

buffers and determine how much of the non-Hispanic white, Hispanic, and black populations of

each county live within the buffers, employing the same technique used in the previous analy-

sis. This is a common empirical strategy in the environmental justice literature, which seeks to

determine which demographic communities are most burdened by the placement of toxic waste

sites (Sheppard et al., 1999; Boer et al., 1997; Centner, Kriesel and Keeler, 1996). Yet, we are

not interested in which communities are burdened by these early voting sites, so much as which

demographic communities have greater access to such an asset. Thus, we calculate the demogra-

phy of the communities that lie within the 1-mile buffers, specifically the percent of the county’s

non-Hispanic whites, blacks, and Hispanics that live within thosebuffers. For instance, if 20,000

Hispanics live within the 1-mile buffers in a county with 100,000 Hispanics, then 20% of the

county’s Hispanics are said to have sufficient access to early voting locations. Given that we ex-

pect Republicans to site more early voting locations in white areas and Democrats to site more in

non-white areas, then we should expect that in counties with Republican county governments, a

greater percentage of the white population will live within the 1-mile buffers relative to non-whites.

For example, a county government comprised of all Republicans might have, say, 24% and 25%

of its black and Hispanic population within the 1-mile buffers, respectively, while 35% of non-

Hispanic whites live within the buffers. Conversely, county governments dominated by Democrats

will have a greater percentage of their non-white populations living within their 1-mile buffers than

their white populations. Thus, we develop two measures as dependent variables—the first is the

difference between the percentage of a county’s non-Hispanic white population and its percentage

12

of Hispanic population lying within the buffers, and the second is the difference between the per-

centage of the non-Hispanic white population and the percentage of black population lying within

the buffers. Positive values for both of these measures would suggest that whites have greater ac-

cess to voting sites within the county, while negative values indicate that non-whites have greater

access to voting sites. For instance, a value of 2 means that 2% more of a non-Hispanic white pop-

ulation has access to early voting sites, relative to the non-white group. We would expect that there

would be a positive relationship between the partisanship of the county and these two measures.

That is, the more conservative the county government, the more whites (percentage-wise) would

live within one mile of an early voting location relative to non-whites. In all subsequent models,

we employ basic OLS models.

Results

As illustrated in Table 2, the degree to which a county government is Republican has a strong and

statistically significant effect on the number of early voting locations, even when controlling for

population. Specifically, for every one additional Republican commissioner or county judge, the

number of early voting locations declines by about 6. More substantively, a county with all Demo-

cratic officials on the county government would have approximately 32 more early voting locations

than a county government composed of all Republican officials. The same holds true in analyses of

full-time and part-time sites. For every additional Republican commissioner or county judge, the

number of full-time early voting locations declines by 1.5, meaning that an all-Democratic county

government would have approximately 8 more full-time sites than a comparable all-Republican

county government. Of the seventeen counties that opted to have temporary or mobile voting loca-

tions, we see a similar relationship, wherein more Republican members on the county government

is correlated with fewer temporary early voting locations. Still, the findings for temporary sites

should be taken with caution, given the small sample size. Ultimately, these results provide some

general support for the Polling Site Quantity Hypothesis, suggesting that Republican county com-

13

Table 2: Effect of County Partisanship on Number of Sites

Dependent variable:

Number of Sites

(All Sites) (Full-Time Sites) (Temporary Sites)

Republican County −6.309∗∗∗ −1.570∗∗ −4.921∗∗

(1.412) (0.591) (1.877)

Total Pop. 0.00001∗∗∗ 0.00001∗∗∗ 0.00001(0.00000) (0.00000) (0.00001)

Constant 36.628∗∗∗ 12.422∗∗∗ 23.443∗∗∗

(6.131) (2.569) (7.444)

Observations 39 39 17R2 0.544 0.761 0.390Adjusted R2 0.519 0.748 0.303Residual Std. Error 13.481 (df = 36) 5.648 (df = 36) 13.309 (df = 14)F Statistic 21.480∗∗∗ (df = 2; 36) 57.441∗∗∗ (df = 2; 36) 4.477∗∗ (df = 2; 14)

Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

missions will create fewer early voting sites in order to increase the costs of voting in pursuit of

more a more favorable electoral environment for their party.

We also seek to assess whether the partisanship of the county government affects the percent

of the total population that lacks access to early voting locations. Below in Table 3, we find

general support for our hypothesis that more of the county’s population will lack access to early

voting sites, as the county government becomes more Republican. Specifically, for every additional

Republican member on the county government, the percent of the population that lives within the

1-mile buffer zones for all early voting sites decreases by 5.2%. That is, a county with an all-

Democratic commission and judge would have 26% more of their population living within the

1-mile buffer zones than a similar county with an all-Republican commission and judge. The same

relationship holds true for full-time sites, wherein for every additional Republican member on the

county government, the percent of the total population living within the 1-mile buffer zones for

full-time early voting locations decreases by 2.2%. In other words, a fully Republican county

government has 10% less of their total population living within one mile of a full-time early voting

14

Table 3: Effect of County Partisanship on Percent Proximate to Early Voting Sites

Dependent variable:

Percent of Population within 1-Mile Bufferl

(All Sites) (Full-Time Sites) (Temporary Sites)

Republican County −0.052∗∗∗ −0.022∗∗ −0.026(0.016) (0.010) (0.031)

County Population −0.00000 −0.00000 0.00000(0.00000) (0.00000) (0.00000)

Area 0.00004 −0.00000 0.0001(0.0001) (0.00004) (0.0001)

Density 0.00001 0.00004 −0.0002(0.0001) (0.0001) (0.001)

Constant 0.417∗∗∗ 0.277∗∗∗ 0.196(0.112) (0.072) (0.159)

Observations 39 39 17R2 0.324 0.166 0.291Adjusted R2 0.244 0.067 0.055Residual Std. Error 0.135 (df = 34) 0.087 (df = 34) 0.166 (df = 12)F Statistic 4.067∗∗∗ (df = 4; 34) 1.686 (df = 4; 34) 1.233 (df = 4; 12)

Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

location, relative to a fully Democratic county government. The results for the temporary sites

are null, but, as noted previously, these results are based on fewer observations and thus may

be unreliable. Again, such results provide some general support for the Polling Site Quantity

Hypothesis, suggesting that Republican county commissions utilize their power to determine early

voting locations in a way that increases the size of population living distantly from the site, thereby

increasing the costs of voting in order to benefit their party’s electoral prospects.

Based on the findings in Table 4, we have little support for our Polling Site Selection Hypoth-

esis. In fact, our results run counter to our hypotheses in the case of black access to early voting

sites. For every additional Republican county official, the percentage of blacks that live within

a 1-mile radius of an early voting location increases by 2% relative to the white population. For

instance, if a county were all Republican, then we would expect that approximately 8% more black

15

Table 4: Effect of County Partisanship on Black Access to Early Voting Sites

Dependent variable:

Percent White Access Relative to Black Access

(All Sites) (Full-Time Sites) (Temporary Sites)

Republican County −0.020∗∗∗ −0.023∗∗∗ 0.002(0.005) (0.005) (0.007)

Constant 0.029 0.048∗∗ −0.028(0.019) (0.018) (0.023)

Observations 39 39 17R2 0.322 0.404 0.004Adjusted R2 0.304 0.388 −0.062Residual Std. Error 0.047 (df = 37) 0.047 (df = 37) 0.049 (df = 15)F Statistic 17.587∗∗∗ (df = 1; 37) 25.093∗∗∗ (df = 1; 37) 0.063 (df = 1; 15)

Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

voters would live within 1 mile of the early voting locations relative to non-Hispanic whites. The

same relationship holds true for the full-time sites. For every additional Republican county of-

ficial, the percentage of black voters living within 1 mile of the early voting locations increases

by 2.3% relative to non-Hispanic whites. While these results seem to contradict our expectations

that, given strong voting correlations between blacks and Democrats, Republican commissions will

place voting sites less proximate to residential concentrations of blacks, we discuss below some

legal reasons why these results run counter to our hypothesis.

Based on the findings in Table 5, we have little support for our Polling Site Selection Hypothesis

with regards to Hispanics. Yet, in this case, the results appear null, suggesting that Hispanics and

whites have relatively equitable access to early voting sites regardless of the partisan affiliation of

the county government. As we discuss below, there are legal reasons to explain the failure of our

demographic hypotheses.

16

Table 5: Effect of County Partisanship on Hispanic Access to Early Voting Sites

Dependent variable:

Percent White Access Relative to Hispanic Access

(All Sites) (Full-Time Sites) (Temporary Sites)

Republican County −0.007 −0.008 0.004(0.006) (0.005) (0.011)

Constant −0.021 −0.012 −0.041(0.023) (0.019) (0.035)

Observations 39 39 17R2 0.036 0.071 0.010Adjusted R2 0.010 0.046 −0.056Residual Std. Error 0.057 (df = 37) 0.049 (df = 37) 0.075 (df = 15)F Statistic 1.393 (df = 1; 37) 2.833 (df = 1; 37) 0.155 (df = 1; 15)

Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

Discussion

Based on the results above, we find support for the theory proposed here. Our results indicate

that Republican county commissions have fewer sites, with less of their total population living

within one mile of their early voting locations. This suggests that Republican county governments

increase the costs of voting for all constituents by having fewer early voting locations. In contrast,

Democrats employ more early voting locations in order to decrease the costs of voting for all

constituents. As we argue above, Republicans seek to depress turnout due to the popular notion

that Republicans win low-turnout elections and Democrats seek to facilitate turnout due to the

notion that Democrats win in high-turnout elections. As good partisan warriors, both Republican

and Democratic county commissioners utilize the prerogatives of their office to bolster their party’s

chances of electoral success across the ballot.

However, there was less obvious support for the notion that county commissions would locate

early voting sites, especially full-time sites, in residential areas made up of favorable racial demo-

graphics. We expected Republican commissions to locate sites geographically distant from blacks

and Hispanics due to voting trends suggesting these demographics tend to vote for Democrats.

17

Similarly, we expected Democratic commissions to locate sites more proximate to black and His-

panic areas. It turns out, Republicans locate sites, especially full-time sites, to be more inclusive of

their county’s black population. The results for Hispanics were largely null, suggesting that county

governments place early voting sites relatively equitably with regard to Hispanics. This may be

a signal that Republicans, or at least those in Texas, make some effort to reach out to minority

demographics. We believe a likely explanation for this finding stems from the Voting Rights Act.

Because Section 2 prohibits any discriminatory practice curtailing the ability to vote on the basis

of race, any attempt to use the prerogatives of the county commission to select voting sites in this

way would be illegal. Therefore, failure to find support for our Polling Site Selection Hypothesis

is not so surprising after all.

Moreover, care must be taken in any discussion of the equity of site selection. That is, are

we discussing equity in terms of all demographic sub-populations having the same access to early

voting sites, or whether the siting of early voting locations contributes to the equitable turnout of

all demographics? For instance, if 34% of blacks, Hispanics, and non-Hispanic whites live within

one mile of an early voting location, but only 26% of Hispanics and blacks vote early, while 40%

of non-Hispanic whites do, is such a situation equitable? As mentioned previously, different racial

and ethnic groups face different costs related to voting. Take, for instance, mobility in Texas. We

can imagine that individuals who have access to their own car in order to drive themselves, as

opposed to walking or taking public transit, to the early voting locations would face lower costs

in voting. Yet, not every racial and ethnic group has equal access to a personal vehicle. In the

counties examined in this analysis, nearly 83% of non-Hispanic whites drive themselves alone to

work, suggesting they have access to their own vehicle and therefore superior means by which

to transport themselves to an early voting location. In contrast, 78% of Hispanics and blacks

drive themselves to work, meaning that 5% less of these sub-populations have access to their own

vehicle. In other words, blacks and Hispanics face greater costs, at least in terms of the mode by

which they arrive at an early voting location, relative to non-Hispanic whites.

The fact that we only find support for one of our hypotheses may be a result of our admittedly

18

limited empirical models. First, Section 2 of the Voting Rights Act is probably a deterrent to using

race as a basis for voting site selection. Second, there is likely more nuance to the idea that Repub-

licans would avoid all minority areas. For instance, there may be a socio-economic component that

we’ve yet to incorporate, like conservative governments targeting low-income white communities

while avoiding low-income minority communities. We hope to delve into such demographic nu-

ances in future iterations. Third, we would also like to analyze how these early voting sites affect

turnout. Currently, we are merely positing that the number and siting of early voting locations

would disenfranchise some groups more than others, depending on the party in power. ? show

that the existence of early voting does not increase turnout. However, they suggest that data on the

quantity and geographic proximity to populations would provide a clearer picture of the effects of

early voting on turnout. We hope to incorporate turnout data into our analysis to test this notion.

Ultimately, our findings that Republican commissions provide fewer early voting locations overall

than Democratic commissions and less of the population in Republican counties fall within a close

distance to the poll than in Democratic counties is encouraging for our theory of local officials as

party operatives.

19

References

Aldrich, John H. 1993. “Rational Choice and Turnout.” American Journal of Political Science

37(1):246–278.

Aldrich, John H. 1995. Why Parties? The Origin and Transformation of Political Parties in

America. The University of Chicago Press, Chicago.

Boer, J. Tom, Manuel Pastor, James L. Sadd and Lori D. Snyder. 1997. “Is There Environmen-

tal Racism? The Demographics of Hazardous Waste in Los Angeles County.” Scial Science

Quarterly 78(4):793–810.

Brady, Henry E. and John E. McNulty. 2011. “Turning Out to Vote: The Costs of Finding and

Getting to the Polling Place.” American Political Science Review 105(1):115–134.

Centner, Terence J., Warren Kriesel and Andrew G. Keeler. 1996. “Environmental Justice and

Toxic Releases: Establishing Evidence of Discriminatory Effect Based on Race and Not In-

come.” Wisconsin Environmental Law Journal 3(2):119–158.

DeNardo, James. 1980. “Turnout and the Vote: The Joke’s on the Democrats.” American Political

Science Review 74(2):406–420.

Downs, Anthony. 1957. An Economic Theory of Democracy. Harper Collins Publishers, New

York.

Dyck, Joshua J. and James G. Gimpel. 2005. “Distance, Turnout, and the Convenience of Voting.”

Social Science Quarterly 86(3):531–548.

Fenno, Richard E. Jr. 1978. Home Style: House Members in Their Districts. Little-Brown, Boston.

Frendreis, John P., James L. Gibson and Laura L. Vertz. 1990. “The Electoral Relevance of Local

Party Organizations.” The American Political Science Review 84(1):225–235.

20

Gomez, Brad T., Thomas G. Hansford and George A. Krause. 2007. “The Republicans Should

Pray for Rain: Weather, Turnout, and Voting in U.S. Presidential Elections.” The Journal of

Politics 69(3):649–663.

Hansford, Thomas G. and Brad T. Gomez. 2010. “Estimating the Electoral Effects of Voter

Turnout.” American Politial Science Review 104(2).

Haspel, Moshe and H. Gibbs Knotts. 2005. “Location, Location, Location: Precinct Placement

and the Costs of Voting.” The Journal of Politics 67(2):560–573.

Highton, Benjamin and Raymond E. Wolfinger. 2001. “The Political Implications of Higher

Turnout.” British Journal of Political Science 31(1):179–223.

Masket, Seth E. 2011. No Middle Ground: How Informal Party Organization Control Nominations

and Polarize Legislatures. The University of Michigan Press, Ann Arbor.

Mayhew, David R. 1974. Congress: The Electoral Connection. Yale University, New Haven, CT.

Rosenstone, Steven J. and John Mark Hansen. 2003. Mobilization, Participation, and Democracy

in America. Longman, New York.

Sanbonmatsu, Kira. 2006. “The Legislative Party and Candidate Recruitment in the American

States.” Party Politics 12(2):233–256.

Schlesinger, Joseph A. 1966. Ambition and Politics: Political Career in the United States. Chicago,

Rand McNally.

Sheppard, Eric, Helga Leitner, Robert B. McMaster and Hongguo Tian. 1999. “GIS-Based Mea-

sures of Environmental Equity: Exploring Thier Sensitivity and Significance.” Journal of Expo-

sure Analysis and Environmental Epidemiology 9:18–28.

Steed, Robert P., John A. Clark, Lewis Bowman and Charles D. Hadley, eds. 1998. Party Organi-

zation and Activism in the American South. The University of Alabama Press, Tuscaloosa.

21

Stein, Robert M. and Greg Vonnahme. 2008. “Engaging the Unengaged Voter: Vote Center and

Voter Turnout.” Journal of Politics 70(2):487–497.

Tucker, Harvey J., Arnold Vedlitz and James DeNardo. 1986. “Does Heavy Turnout Help

Democrats in Presidential Elections?” The American Political Science Review 80(4):1291–1304.

22

Notes

1The states that allow early voting are: Alaska, Arizona, Arkansas, California, Colorado, D.C., Florida, Georgia,

Hawaii, Idaho, Illinois, Indiana, Iowa, Kansas, Louisiana, Maine, Maryland, Montana, Nebraska, Nevada, New Mex-

ico, North Carolina, North Dakota, Ohio, Oklahoma, South Dakota, Tennessee, Texas, Utah, Vermont, West Virginia,

Wisconsin, Wyoming

2These data were collected from the National Conference of State Legislatures at http://www.ncsl.org/research/elections-

and-campaigns/absentee-and-early-voting.aspx

3http://www.orangecountync.gov/departments/boardofelections/index.php

4These data were collected by the authors using ballotpedia.org, local media, county government websites, and

secretary of state websites.

5These data were obtained from the National Association of Counties at http://www.naco.org/sites/default/files/documents/County%20Elections%20Partisian%20or%20Non%20Partisian.pdf.

6Tennessee uses county Election Commissions appointed by the state election commission to select locations.

7Texas Election Code: http://county.org/member-services/legal-resources/publications/Documents/Guide-to-Laws-

for-County-Officials.pdf

8United States Election Assistance Commission: http://www.eac.gov/assets/1/VirtualMeetingRoom/AVM.doc

9Texas Election Code: http://www.statutes.legis.state.tx.us/Docs/EL/htm/EL.85.htm

10For example, see here: http://mctx.org/dept/departmentsc/commissionerscourt/.

11Schlesinger also mentions discretely ambitious politicians who seek no reelection. We assume these politicians

are few and are generally party loyalists.

12These statistics were obtained from the Pew Research Center’s 2014 Religious Landscape Study (RLS-II).

23