University of Southern California Dissertations and Theses

Investigating electoral college reform: geography's impact on elections, and how maps influence our perception of election outcomes

(USC Thesis Other)

Investigating electoral college reform: geography's impact on elections, and how maps influence our perception of election outcomes

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

INVESTIGATING ELECTORAL COLLEGE REFORM:
GEOGRAPHY’S IMPACT ON ELECTIONS, AND HOW MAPS INFLUENCE OUR
PERCEPTION OF ELECTION OUTCOMES

By

Meagan Calahan

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)

August 2014

Copyright 2014 Meagan Calahan

ii

ACKNOWLEDGEMENTS
I would like to express my gratitude to my committee chair, Dr. Karen Kemp, for her
guidance and for her endless supply of patience with me while trying to determine my
approach for this topic. Thanks also to the other members of my committee, Dr. Robert
Vos and Dr. Daniel Warshawsky, both of whom took time out of their busy schedules to
provide support to me on this project when I needed it.
A special thank you to my colleagues and unofficial mentors, Matt Price and
Gulla Gisladottir, for believing in me enough to encourage me to pursue my Master’s
degree—and for reminding me why it was worth it when I got discouraged.
Last but not least, huge thanks to my family—especially my husband, Mike—for
their love and support, and for never complaining when I spent all of my nights and
weekends focused on school.

iii

TABLE OF CONTENTS

Acknowledgements ii
List of Tables iv
List of Figures v
Abstract vii
Chapter 1: Introduction 1
Chapter 2: Background 4
2.1 Controversial Elections and Other Problems 6
2.2 Proposed Reform Methods 8

Chapter 3: Data and Preliminary Analysis 12
3.1 Election Results 16
3.2 Effect of the Proposed Reform Methods 19

Chapter 4: Mapping Data 21
4.1 Fundamental Mapping Principles 23
4.2 Overview of Common Thematic Mapping Techniques 24

Chapter 5: Mapping the Election Results 31
5.1 Mapping the Traditional Electoral College Method 32
5.2 Mapping the Popular Vote Method 37
5.3 Mapping the Proportional Allocation Method 44
5.4 Mapping the Congressional District Method 48

Chapter 6: Conclusions 52

References 55

iv

LIST OF TABLES

Table 1: Congressional District dataset attribute table before join 15
Table 2: Popular Vote Results for the 2012 Presidential Election 17
Table 3: Proportional Allocation Results for the 2012 Presidential Election 18
Table 4: Congressional District Results for the 2012 Presidential Election 19
Table 5: Results for 2000, 2008 & 2012 Elections for all Reform Methods 19

Table 6: Comparison of Techniques for Mapping Election Results 32

v

LIST OF FIGURES

Figure 1: Electoral College results for the 2012 Presidential Election 22

Figure 2: Voting results of the 2012 election in Los Angeles County, California,
in which a single point represents each vote 27

Figure 3: Voting results of the 2012 election in Los Angeles County, California,
in which a single point represents 1,000 votes 28

Figure 4: Electoral College results for the 2012 Presidential Election,
presented in a Choropleth format based on the amount of electoral votes
awarded by each state 33

Figure 5: Electoral College results for the 2012 Presidential Election,
presented in a combination choropleth and graduated symbol format 34

Figure 6: Electoral College results for the 2012 Presidential Election,
presented as a cartogram 35

Figure 7: Electoral College results for the 2012 Presidential Election,
presented as a cartogram with the outlines of the United States as an overlay
to demonstrate the extent to which the geography has been altered 36

Figure 8: Voting results of the 2012 election nationwide,
in which a single point represents 10,000 votes 39

Figure 9: Voting results of the 2012 election nationwide,
in which a single point represents 50,000 votes 40

Figure 10: Republican Party voting results of the 2012 election nationwide,
in which a single point represents 50,000 votes 41

Figure 11: Democratic Party voting results of the 2012 election nationwide,
in which a single point represents 50,000 votes 42

Figure 12: Proportional symbol map of the 2012 election,
using the Proportional Allocation method 45

Figure 13: Graduated symbol map of the 2012 election,
using the Proportional Allocation method 46

Figure 14: Bar chart map of the 2012 election,
using the Proportional Allocation method 47

vi

Figure 15: Congressional District results for the 2012 election,
minus the two Senatorial votes per state 49

Figure 16: Congressional District results for the 2012 election,
with state-level overlay 50

vii

ABSTRACT

Multiple events throughout the history of the United States of America have led people to
call for the Electoral College system to be reformed or abandoned altogether. As the
Electoral College currently functions, each state awards a set number of votes
(determined by population) to the candidate who receives the largest number of votes, but
many citizens feel that there are flaws in this system. Although there have been many
reform propositions over the years, there are three potential methods that consistently
have the most support: Popular Vote, Proportional Allocation, and Congressional
Districts Allocation. This study offers insight into how each of these reform methods
might change election outcomes and even more importantly, by exploring several
possible election mapping techniques, it provides an analysis of how the presentation of
election results in a geographic format can alter the viewer’s perceptions of election
outcomes and of the viability of the various reform methods. Finally, this study provides
arguments for why the traditional methods of representing election outcomes tend to fall
short.

1

CHAPTER 1: INTRODUCTION
Throughout American history, there has been much debate about the efficacy of the
method of electing the President and Vice President of the United States—most of which
is centered around the fact that it is not necessary for a candidate to win the popular vote
in order to become President. Because of this fact, presidential campaigns tend to focus
their time and money strategically; the candidates inundate battleground states and
undecided voters with campaign ads and personal appearances, while traditionally
Republican or Democratic states and citizens are largely ignored.
This debate intensified following the election of 2000, which had the distinction
of not only being the only election in recent history in which the winner of the popular
vote was not elected President (the election outcome went all the way to the Supreme
Court, and the winner was ultimately decided by a 5-4 vote), but was also the first
election since 1864 to have a faithless elector (the elector for Washington, D.C., Barbara
Lett-Simmons, refrained from voting to protest the fact that D.C. has no representation in
Congress) (Schultz 2009). These events naturally led to a renewed interest in reforming
the Electoral College, with several camps lobbying for their preferred reform method to
replace what they view as a flawed system.
Over a decade after that last controversial election, the debate over the necessity
and manner of Electoral College reform continues. Unfortunately, for most of the voting
public, the Electoral College and the methods being proposed to replace it are still
something of a mystery. This study was created in order to gain a better understanding of
the subject, by approaching the analysis of each proposed method from a geographic
perspective. By doing so, it attempts to answer three key questions:

2

1. Does changing the geography of vote calculation in Presidential Elections have
an effect on the election outcomes? Each of the proposed Electoral College reform
methods uses a different geographic boundary by which votes are calculated and reported.
The Proportional Allocation method awards Electoral College votes at the state level, the
Congressional District method awards Electoral College votes at the Congressional
District level, and the Popular Vote method simply aggregates votes at the national level.
In theory, the fact that all of the calculations are based on the same voting outcomes, the
election outcomes should all be the same. However, the arguments about which reform
method is superior lead to the supposition that this is not actually the case.
2. Given the unique and varied geography of the United States, is it possible to
create maps that clearly convey information not only about the election outcomes when
each reform method is used, but also about the location and voting behavior of United
States citizens? Traditional election maps produced by the mass media typically exist
solely to illustrate which candidate won each state’s electoral votes, failing to take into
account the enormous impact that voter population and the available electoral votes have
on the final election outcome. This often results in confusion due to the disparity in both
geography and population size between the states, but the election mapping standard
remains the same. The wide array of symbology available in GIS offers us a chance to
explore other methods of representing election outcomes in a way that is less confusing to
the viewers.
3. What do these maps actually tell us about how well each of these different
methods really represents the will of the people? As is outlined in Chapter 2, each of the
reform methods has well-known pros and cons. Can we use the spatial arrangement of

3

voting Americans as they are spread across the country, illustrated in maps, to help us
understand how well each of these reform measures reflects how individuals have voted?
This study begins by investigating why the Electoral College was created in the
first place, and what our founding fathers hoped it would achieve. It goes on to explore
where the process began to fail in the eyes of those who oppose it, why many voters feel
that it is a flawed system, and what alternatives have been suggested. In order to
determine whether or not the proposed reform methods change the outcome of the
elections in a manner that is a better representation of the voters, each method is applied
to two past presidential elections: the controversial election of 2000, and the most recent
election of 2012.
Finally, an analysis of the challenges inherent in mapping these results is
undertaken, by exploring different methods of creating maps of the results of the 2012
election using different mapping techniques. The mapping exploration offers insight into
how changing the visual representation of the data changes the way the viewer interprets
the data. Further, it examines the impact that the geography of vote calculation has on
election outcomes—exploring some of the claims made by those who are in favor of
Electoral College reform in order to determine which claims have validity and which
should be disregarded in the future. Lastly, it illustrates the process of creating a map
that accurately represents the voters and election outcomes visually, focusing on the
benefits and shortcomings of some of the common (and one uncommon) types of
symbology available with Esri’s ArcMap software.

4

CHAPTER 2: BACKGROUND
One of the many topics up for debate during the Constitutional Convention of 1787 was
the manner by which the new country would choose its Presidents. The most popular two
methods proposed at that time were direct election by Congress and direct election by the
people, but both were met with strenuous opposition. Those who opposed the direct
election by Congress argued that it would deny the President autonomy, as he would feel
the need to be subservient to Congress in order to secure re-election. Opposing arguments
to direct election by the people included the fear of a lack of influence on the elections by
both smaller states and Southern states with high non-voting slave populations, as well as
a fear that the lack of knowledge of the candidates from state to state would lead to
people only voting for local favorites (which would most likely lead to only people from
populous areas being elected) (Belenky 2013).
As a compromise to both sides, the third Committee of Eleven ultimately settled
on an indirect election approach (Vile 2005). Although it was actually the second choice
of many of the delegates, this method—involving a college of electors—was designed as
a compromise in order to eliminate the fear of under-representation for the smaller and/or
Southern states by giving each state one elector per Congressional representative and one
for each Senator (Rose 1994). States had the option of whether or not to hold a popular
vote election--despite what many people believe, holding a popular vote election is not
Constitutionally mandated, and in fact South Carolina abstained from holding one until
after the Civil War, for the election of 1868 (Levy 2009). Once this compromise was
proposed, the Electoral College method of electing a President was subsequently ratified
with little contention or debate (Longley and Peirce 1996).

5

At its inception, the Electoral College was structured slightly differently than it is
today. The original structure called for each state to receive one elector for each of its
Senate members (totaling two per state), and one elector per member of the House of
Representatives (which varies based on population, but is never less than one)—Senators
and Representatives were barred from serving as electors, however. Electors were to
meet in their respective states to cast their vote rather than meeting in a more central
location, and were expected to vote for two Presidential candidates, with no vote for Vice
President. When the votes were counted, the candidate receiving the Electoral majority
would be elected President, and the candidate who came in second place would be
awarded the Vice Presidency. If no Electoral majority was reached, the House of
Representatives would step in to decide the winner (U.S. Constitution, Art. II Sec. I).
While the basic structure of the Electoral College still remains the same, there
have been several changes over the years—electors changed from being expected to
choose who they felt was the best candidate to being pledged to represent a particular
political party (which is the method predominantly used today), states almost
unanimously adopted a “winner take all” system of awarding electoral votes, and a clear
two-party system emerged (Longley and Peirce 1996). Perhaps most notable, though, was
the change that led to the 12
th
Amendment to the Constitution. This Amendment
abandoned the previous two-vote method of choosing President and Vice President and
replaced it with a system in which electors cast one vote for a President and Vice
President running on the same ticket (U.S. Constitution, Amend. XII).

6

2.1 Controversial Elections and Other Problems
For the most part, the Electoral College system has worked well over the course
of our nation’s history. As intended by the authors of the Constitution, it has succeeded
in giving small states a voice in Presidential elections, which many argue contributes to a
more unified country (Kimberling 1992). In fact, the vast majority of elections have gone
smoothly and without dissent.
However, in four different elections (1824, 1876, 1888 and 2000), the winner of
the popular vote failed to win a sufficient majority of electoral votes and subsequently
lost the election. This roughly 7% discrepancy rate, among other factors, has led to
widespread debate about the efficacy of the Electoral College since very shortly after its
inception (Jenkins and Sala 1998). Opponents of the system argue that it is not a fair
representation of the will of the majority of the voters—a candidate only needs to win
51% of the popular vote in a state to win 100% of the electoral votes, which has the
potential to skew election outcomes (as in the aforementioned elections). Perhaps as a
result of this, a large number of states, including the three most populous states,
consistently distribute their electoral votes to the same party and have come to be
considered “safe” for either Democratic or Republican candidates.
This fact not only contributes to lower voter participation in those areas—voter
turnout in safe states is typically five to ten points lower than in swing states—it also
means that candidates feel that they don’t have to focus their attention on campaigning in
these states, and instead choose to focus both their time and money on the swing states
(Deschamps et al. 2012, Black 2012).

7

A swing state is defined as “a US state where the two major political parties have
similar levels of support among voters, viewed as important in determining the overall
result of a presidential election” (Oxford Dictionaries 2013, no page). These states are so
important to political campaigns, in fact, that they have become the primary focus of the
candidates in recent years.
Swing states have a pivotal impact on how campaigns are focused, both
financially and strategically. Because roughly 40 states are considered a lock for one
candidate or the other before the election ever starts, they are historically almost
completely ignored during campaign season; typically, they receive no special TV ads or
campaign visits (aside from the occasional fundraiser). Instead, all campaign stops and
targeted TV ads (in other words, the vast majority of the campaign financing) are
reserved for the battleground or swing states, where the candidates hope to win undecided
voters over to their side.
Perhaps more troubling than the lopsided spending, though, is the fact that this
extreme focus on the opinions of swing states also creates lopsided campaigning. In
trying to please the swing state voters, candidates have been known to assign extra
importance to the needs of the residents in those states in order to win their votes (such as
Medicare to secure the votes of the elderly in Florida). This can also lead to similar
behavior on the part of a first-term President hoping to win a second term, because they
are more likely to make executive decisions that are skewed by the knowledge of what
will win or lose votes from the swing states in the next election (Black 2012).

8

2.2 Proposed Reform Methods
While there have been many Electoral College reform methods proposed over the
years, the three that are offered up most consistently are the focus of this study: Popular
Vote, Proportional Allocation, and Congressional District Allocation.
Popular Vote
The Popular Vote method proposes that the Electoral College be eliminated
altogether, in favor of a National Popular Vote. This method has the distinction of having
come very close to becoming a reality: following the 1968 Presidential election (where
Richard Nixon won 51% of the popular vote but won 56% of the electoral votes), House
Joint Resolution 681 was created in favor of amending the Constitution to award the
Presidency to whichever candidate won at least 40% of the popular vote. If no candidate
won by that margin, or if there was a tie, a runoff election would be conducted between
the two candidates who had earned the highest number of votes. This Resolution passed
in the House of Representatives and was endorsed by President Nixon, but was subject to
filibuster in the Senate and was ultimately abandoned (Johnson 2009).
Proponents of this method feel that it is superior to the Electoral College because
it better represents the will of the people, and would eliminate the possibility of the less
popular candidate winning the election. It would also eliminate the possibility of swing
states, as states would no longer carry any electoral votes—thus, campaigns would
theoretically become less lopsided and more focused on the entire population, rather than
on the desires of the swing voters (Anderson 2001).
Opponents of this method fear that the lack of swing states would lead to
significantly more expensive elections, as candidates would find it necessary to campaign

9

to the entire country instead of just to the undecided voters in key states. Additionally,
many argue that a popular vote could weaken the power of the states (especially those
that are less populous) (Belenky 2008).
Proportional Allocation
Unlike the National Popular Vote, the Proportional Allocation method would not
require a Constitutional amendment in order to be implemented, because it does not
eliminate the Electoral College altogether. Instead, it changes the manner in which the
electoral votes are awarded—rather than awarding electoral votes on a winner take all
basis, it would award the electoral votes proportional to the popular vote. In other words,
if a candidate won 60% of the popular vote in Texas, for example, he or she would be
awarded 60% of Texas’ electoral votes (Neale and Whitaker 2004).
Proponents of this method feel that it has the dual benefits of better representing
the popular vote while also staying true to the intentions of the Constitution.
Additionally, it would create a more “national” election, as no one state would carry a
guaranteed amount of electoral votes.
Those who oppose this method feel that it would undermine the current two-party
system (since third parties could potentially win electoral votes without winning the state),
and that it would take away the power of the smaller states in elections because they
would have potentially fewer electoral votes to offer (Neale 2009)
Congressional Districts
Like the Proportional Allocation method, the Congressional District method has
the benefit of not requiring a Constitutional Amendment in order to be implemented, as
the manner in which electoral votes are awarded is decided by the states. In this method,

10

one electoral vote is awarded to the winner of each Congressional district, with the
remaining two votes being awarded to the winner of the overall majority for that state
(The Center for Voting and Democracy 2009).
Proponents of this method feel that it has many of the same benefits of the
Proportional Allocation method, in that it is a better representation of the popular vote
while still remaining true to the Constitution—and, since Maine and Nebraska have
awarded votes this way since 1972 and 1992 respectively, it comes with a proven history
of success (The Center for Voting and Democracy 2009).
Unfortunately, the prevalence of Gerrymandering raises concerns for many people.
Gerrymandering is the practice of redrawing electoral district boundaries for the specific
purpose of increasing the advantage for a certain political party (Koehler 2010). Because
this practice is common in Congressional Districts, many feel that the distribution of
electoral votes would potentially be skewed, giving one political party an unfair
advantage. (Hirsch 2008) Additionally, many feel that rather than fixing the swing state
problem, it would shift the focus of campaigns from undecided states to undecided
districts (The Center for Voting and Democracy 2009).
While each reform method clearly has both strong supporters and strong
opposition, it is important to set aside political ideology and answer one key question:
Does this reform method accurately represent the wishes of the voters? If it does not,
then it is no better than the system that is already in place. If it does, then perhaps it is
worth exploring further. An exploration of past election results, re-calculated according
to the top three reform propositions can help to answer that question. In order to begin

11

such an examination, it is necessary to find and explore election data for more than one
election to determine how each reform method might impact the outcomes.

12

CHAPTER 3: ELECTION DATA AND PRELIMINARY ANALYSIS
Coming up with an effective way to analyze Electoral College reform was surprisingly
difficult, and this study changed direction several times while trying to determine an
approach. Initially, the intention was to study the voting behavior of certain demographic
groups and use population projections provided by the U.S. Census Bureau to predict and
compare future election outcomes using each of the proposed reform methods. After
many unsuccessful attempts at using regression analysis to predict the voting behavior of
specific demographic groups, it became clear that this approach would not work, for two
primary reasons: one, no one person can be defined as a single demographic (which was
the reason regression analysis wouldn’t work); and two, predicting future elections is
incredibly difficult. There is an entire industry of professionals who attempt to predict
election outcomes, and even those professionals find the science to be uncertain.
Thus, it was ultimately decided that the research questions that were to be answered
didn’t require a look into the future—they could be answered just as well by studying
past elections. Those key questions were: how did changing the geography by which the
votes are counted change the way the voice of the voters is represented and, how do maps
of the election outcomes help us understand this question of representation?
Because this is a study of elections taking place in the United States, it was
obvious that all data downloaded would be focused in North America, but it was less
obvious at what resolution the data should be displayed. Representing all election results
on a National level was dismissed immediately as being too broad, because it did not
allow exploration of the geographic changes that each reform method suggested.
Unfortunately, there was no one perfect choice that would work for every reform method,

13

which led to the decision to use United States data at three different levels: state, county
and congressional district.
The state level boundaries were downloaded from the National Atlas website and
were used to map the traditional Electoral College results, as well as the results when
using the Proportional Allocation method (because that method also awards Electoral
votes at a state level). The county level boundaries were also downloaded from the
National Atlas website and were not strictly necessary, since none of the proposed reform
methods award votes on a county level. However, the data was useful when mapping the
Popular Vote results as they better illustrate the location of densely populated areas,
which has a strong correlation with voting behavior. The 106
th
and 112
th
congressional
district polygon datasets were of course essential to illustrate election results using the
Congressional District method, and were added via ArcGIS online, which compiled the
data from the U.S. Census TIGER files. In order to maintain the correct shape of the
geography, three different projections were used: the contiguous United States were set
to Lambert Conformal Conic projection; Hawaii was set to Albers Equal Area Conic for
Hawaii; Alaska was set to Albers Equal Area Conic for Alaska. Positional accuracy was
not a necessary component of this study, and so it was not factored in.
Once the boundaries were downloaded, the next step in the process was to find
election data for the 2000 and 2012 Presidential elections at the national, state and
congressional district level. Election results are calculated at the precinct level and are
aggregated to determine the state and national results. While all of the resulting tables
could have been aggregated from the original precinct-level data, the data was easy to
find at each level, which eliminated the need for extra table manipulation. All of the

14

tabular data was found at the US Census website, although it required some manipulation
(detailed in the next section) to determine the election outcomes with the Proportional
Allocation and Congressional District methods, and a small amount of formatting before
any of the tables could be joined with the boundary data.
The attribute tables of each of the boundary datasets came with extensive
demographic information about the states, counties, and congressional districts from the
2000 and 2010 censuses (including population, population density, household size,
income, age, race, etc.), but did not have any election information (Table 1). To make it
possible to create maps of the election results, then, a join was created between the
election results data and the polygon datasets on the FIPS field. This created a larger
table for each dataset that included the election results along with demographic data, and
made creating a map of those results possible. It also created a very unwieldy table, with
quite a bit of unnecessary information. To trim it down to a more manageable size, all of
the extraneous demographic information was removed from the tables, until they
consisted of only the information that pertained to this study. Once all of the boundary
data was in place and the tables were properly formulated, it was time to start the analysis.

15

Table 1: Congressional District dataset attribute table before being joined with election
results

OBJECTID DISTRICTID STFIPS CD113FIPS STATE NAME LAST_NAME PARTY
1 613 6 13 CA Barbara Lee Lee Democrat
2 1501 15 1 HI Colleen W. Hanabusa Hanabusa Democrat
3 4821 48 21 TX Lamar Smith Smith Republican
4 4828 48 28 TX Henry Cuellar Cuellar Democrat
5 4810 48 10 TX Michael T. McCaul McCaul Republican
6 4815 48 15 TX Rub√©n Hinojosa Hinojosa Democrat
7 4820 48 20 TX Joaquin Castro Castro Democrat
8 4835 48 35 TX Lloyd Doggett Doggett Democrat
9 4822 48 22 TX Pete Olson Olson Republican
10 4809 48 9 TX Al Green Green Democrat
11 4829 48 29 TX Gene Green Green Democrat
12 4807 48 7 TX John Abney Culberson Culberson Republican
13 4818 48 18 TX Sheila Jackson Lee Lee Democrat
14 4802 48 2 TX Ted Poe Poe Republican
15 1212 12 12 FL Gus M. Bilirakis Bilirakis Republican
16 1206 12 6 FL Ron DeSantis DeSantis Republican
17 1217 12 17 FL Thomas J. Rooney Rooney Republican
18 1205 12 5 FL Corrine Brown Brown Democrat
19 1210 12 10 FL Daniel Webster Webster Republican
20 1215 12 15 FL Dennis A. Ross Ross Republican
21 1209 12 9 FL Alan Grayson Grayson Democrat
22 1211 12 11 FL Richard B. Nugent Nugent Republican
23 1224 12 24 FL Frederica S. Wilson Wilson Democrat
24 1223 12 23 FL Debbie Wasserman Schultz Schultz Democrat
25 1225 12 25 FL Mario Diaz-Balart Diaz-Balart Republican
26 1208 12 8 FL Bill Posey Posey Republican
27 1218 12 18 FL Patrick Murphy Murphy Democrat
28 1220 12 20 FL Alcee L. Hastings Hastings Democrat
29 1221 12 21 FL Theodore E. Deutch Deutch Democrat
30 1222 12 22 FL Lois Frankel Frankel Democrat
31 1216 12 16 FL Vern Buchanan Buchanan Republican
32 1207 12 7 FL John L. Mica Mica Republican
33 4804 48 4 TX Ralph M. Hall Ralph M. Hall Republican
34 3202 32 2 NV Mark E. Amodei Amodei Republican
35 616 6 16 CA Jim Costa Costa Democrat

16

3.1 Election Results

Determining the results for the two elections under different reform methods was
simply a matter of creating a series of tables—each of which divided up the actual
election results based on one of the proposed Electoral College reform methods—and
adding up the columns for each party in order to determine the winner. In the case of the
Popular Vote (Table 2), this process was straightforward addition; other reform methods
required another step (or two) of calculations.
The Proportional Allocation table (Table 3) used the popular vote data to calculate
the percentage of the total votes per state that went to the Democratic candidate, and then
applied that percentage to the allotted Electoral College votes per state in order to
determine how many Electoral votes were awarded to each candidate.
Because the Congressional District Allocation method gives one Electoral vote to
the winner of each district in a state and two votes to the winner of the most districts, a
table was created that calculated the winner of each district, assigned Electoral votes
accordingly, and then added the total number of Electoral votes per candidate in order to
determine the winner of the two Senatorial votes (Table 4).

17

Table 2: Popular Vote Results for the 2012 Presidential Election

18

Table 3: Proportional Allocation Results for the 2012 Presidential Election

19

Table 4: Congressional District Results (portion) for the 2012 Presidential Election

3.2 Effect of the Proposed Reform Methods
The results of each individual table were then compiled into one table, where they
were compared with the original Electoral votes for each election in order to understand
the impact of each reform method (Table 5).

Table 5: Results for 2000 & 2012 Elections for all Reform Methods
(winner shown in
bold)

*Note: The elector for Washington D.C. abstained from voting in the 2000 election,
becoming the first faithless elector since 1864

20

The table makes it clear that each proposed reform method has the potential to
change the outcome of the election, with the controversial 2000 election showing a large
degree of variation depending on the calculation method, and the Congressional District
Allocation method showing more variation than in other proposed reform methods which
could possibly be due to the effects of Gerrymandering. Popular Vote and the
Proportional Allocation method are the most consistent, as they are not based on
geographic boundaries or a winner-take-all approach, but instead represent the voting
public directly.
As mentioned in the Introduction, these tables give us valuable information, but
there are still many questions to be answered, particularly about the pros and cons that
have been used in the arguments for and against each of the proposed reform methods.
Because finding the answers to these questions goes beyond the scope of a simple table,
we now turn to a consideration of how this data can be represented on maps. The next
chapter introduces a few essential cartographic principles, and then Chapter 5 applies
these principles to the election results discussed here.

21

CHAPTER 4: MAPPING DATA
If tables can answer the question of who the winner of any past election would be with
each of the proposed Electoral College reform methods, can maps provide any additional
understanding? Technically, maps are not needed to provide the results of the simulation
of different reform proposals —a map will not contribute any new information about who
won or lost the election, no matter which reform method you choose. However, maps
still contribute a great deal to the conversation by providing a visual assessment of where
the voting population resides and how changing the geography of the vote calculation can
enhance or silence the voice of voters from different regions.
Additionally, whether it is done consciously or unconsciously, maps can be
manipulated to provoke a variety of responses in the viewer—and to tell many different
stories—without ever changing the data. Because of this, it is very important that anyone
viewing an election results map (or a map of anything, for that matter), understand that
every map manipulates reality to a certain extent, coloring one’s perception of the data
(Monmonier 1996). Creating maps of the tabular election results is not as simple as
putting the data into a geographic context. In traditional depictions of the Electoral
College results (Figure 1), each state is colored either red or blue in order to easily
identify the winner of the Electoral votes for that state (for an example of this, see
Gelman 2014), although it is only since the 2000 election that red was officially the color
of the Republican candidate and blue the color of the Democrats (Enda 2012).

22

Figure 1: Electoral College results for the 2012 Presidential Election

While this approach is technically correct, it is visually misleading. To the casual
viewer, the country seems dominated by the larger red states in the southern and mid-
western regions, while the smaller blue states are easily overlooked. This distribution of
red and blue leads many to believe that the Republican candidate won the election, when
in fact the Democratic candidate, Presidential Incumbent Barack Obama, was the winner
by a fairly large margin.
This raises the question: how can a map more accurately reflect the election
results in a country with such vast variations in state size and population? There are
many possible approaches, but all of them require sacrificing some information in order
to more accurately represent others.

23

4.1. Fundamental Cartographic Principles
There are fundamental cartographic principles that must be observed when
making any map, to ensure that the data is both represented clearly and is an accurate
depiction of the phenomena that are most important. In order to ensure that these criteria
are met, every map—no matter what the subject—should be created with two basic goals
in mind: The first is to make something that conveys the intended information in a way
that is useful to the map user, and the second is to create a map that communicates
information in the most efficient and clear manner possible (Slocum, et al. 2009).
Slocum et al. suggest that every map that is created with these goals in mind should be
made with the following components, listed in order of importance (from most to least
important):
• Thematic symbols and labels that directly relate to the theme of the map
• A clear title, subtitle, and legend
• Base data—such as boundaries, roads and city names
• A scale bar (or text) and north arrow
• Notes regarding data and/or sources
• A frame and neat lines
Once those basic guidelines are met, the more complex components of map
creation can be considered. The cartographer must examine the data and decide what
story the map is telling, what information is most important to the telling of that story,
and how that information can be communicated effectively.

24

4.2 Overview of Common Thematic Mapping Techniques
Before the process of mapping the election results using each of the proposed
Electoral College reform methods begins, it is important to gain a general understanding
of different thematic mapping techniques, what type of phenomena they best represent,
and some potential shortcomings for each. Thematic maps are designed to illustrate a
particular theme (or themes) over a specific geographic area, and are the focus of this
study. The following section provides a brief introduction to the techniques that were
used within the context of this study, while the actual maps of the results are explored in
Chapter 5.

Choropleth Maps
“A map in which enumeration units (or data-collection units) are shaded
with an intensity proportional to the data values associated with those
units” (Slocum et al. 2009, p. 502).

Choropleth mapping is primarily used for phenomena that have a spatial variation
that is the same as the boundaries being used in the study (county, state and congressional
districts in this case). This type of mapping involves assigning a graduated color scheme
to different classes of numeric data, generally with light colors representing low values
and dark colors representing high values. In the case of this study, the choropleth
technique is particularly useful to convey the differences in the number of Electoral votes
awarded per state.
Choropleth maps are used liberally to demonstrate many national statistics at the
state (and often county) level, but they are not without their problems. Primary among
them is the modifiable areal unit problem (MAUP), which is a very common issue that
occurs when performing analysis on aggregated data. The MAUP can take one of two

25

forms: the scale effect (which is exhibited when data is aggregated to different sizes of
zones, such as counties and states) and the zone effect (which is exhibited when the
general size of the zones remains constant but the boundaries of the zones are changed)
(Manley 2014). These two effects can significantly change the statistics resulting from
tabular or spatial analysis. While this study does not involve any in-depth spatial analysis,
the MAUP still presents a problem for map audiences performing a basic visual analysis
of the results—the huge amount of areal variation that occurs within the political
boundaries of the United States creates a different impression of the results of the election
depending on the level of aggregation presented (state, county or congressional district),
creating both the zone and scale effects.
Choropleth maps do offer one way to mitigate the misleading visual zone effect of
the usual blue/red binary election map shown in Figure 1. By using graduated colors it is
possible to better represent the weight (or lack thereof) that each state’s Electoral votes
carry. This has the benefit of de-emphasizing the apparent impact that some of the larger
but less populous states have on the election outcome, however, as shown in the next
chapter the map still ultimately suffers from the visual weight that the larger states carry.

26

Proportional Symbols
“A map in which point symbols are scaled in proportion to the magnitude
of data occurring at point locations” (Slocum et al. 2009, p. 513).

Like choropleth maps, proportional symbols are typically used to represent
numerical data that is associated with geographic locations. Often, this data is measured
at and associated with specific point locations such as cities or work site locations,
though proportional symbols can also be used for data that is collected over areas but
represented as centrally located points. In the case of this study, the latter approach was
used; proportional symbols were created to visually represent the portion of the Electoral
votes that each state would have granted using the Proportional Allocation method. Using
proportional symbols, the division of votes within a state can be shown without altering
that state’s geography. After all—a state’s geography cannot be bisected simply because
it would make symbolization more convenient.
This thematic mapping technique is a simple and effective method of conveying
information with large size disparities, which is why it was a useful tool for mapping the
traditional Proportional Allocation results, but is often not detailed enough to convey the
difference between two similar values, as will be demonstrated in Chapter 5.

Dot Density Maps
“A map in which small symbols of uniform size (typically, solid circles) are used
to emphasize the spatial pattern of a phenomenon” (Slocum et al. 2009, p. 505).

With the dot mapping technique, one dot is placed within a spatial context to
represent a set number of a particular phenomenon—in the case of this study, one dot can
be used to represent a given number of votes to illustrate the popular vote outcome.

27

Choosing the number of entities to assign to a single dot is critical. Trying to
represent each individual as a single dot is impractical—even keeping the symbology as
small as possible, things quickly overlap and become confusing. For example, Los
Angeles County, California is 4,752 square miles and has a population of roughly 10
million people—over 2 million of who voted in the last election. When the results of the
2012 election are represented with one tiny dot for each person who voted, each colored
blue or red to indicate the party of their vote, the map is impossible to interpret (Figure 2).
Figure 2: Voting results of the 2012 election in Los Angeles County, California, in which
a single point represents each vote.
Note: Dots are distributed randomly

28

In order to make a map that actually conveys information clearly, it is necessary
to sacrifice some of the accuracy of representing every vote, in favor of aggregation.
Using the Los Angeles County example again, this time with one dot representing 1,000
votes, things quickly become easier to understand (Figure 3). There is still no way to
determine at a glance exactly how many votes each candidate received in the election, but
the revised map makes it more clear that blue is the predominant color, indicating that the
Democratic candidate won the majority of the votes (which is in keeping with the actual
results for the county, where Barack Obama won by almost one million votes).
Figure 3: Voting results of the 2012 election in Los Angeles County, California, in
which a single point represents 1,000 votes.
Note: The lack of randomly distributed blue dots in the western portion appears to be an artifact of the
dot placement algorithm, possibly related to memory allotment. While visually distracting, the number
of dots overall is correct.
Kern
County
Los Angeles
County
Ventura
County
San
Bernardino
County
Orange
Los Angeles County Popular Vote - 2012 Election
Los Angeles County
1 Dot = 1,000
OBAMA
ROMNEY
Neighboring Counties
0 7.5 15 22.5 30 3.75
Miles

29

Cartograms
“A map that purposely distorts geographic space based on the values of a theme
(e.g., making the size of countries proportional to their population)” (Slocum et al.
2009, p.502).

Cartograms were created as a solution to the areal problems that are so often
encountered when attempting to represent data about different geographic regions. In
this method, geographic regions are distorted to represent the weight of a thematic
variable (such as population, average income, or a particular health issue). Thematic
variables are given visual importance at the expense of preserving geography. Though
the representation can vary, there are two traditional types of cartograms: contiguous
(which maintain the geographic contiguity of the areas being represented but the shapes
are distorted) and noncontiguous (which retain the shape of the areas being represented at
the expense of contiguity) (Slocum et al. 2009).
Creating cartograms without the aid of computers involves complex mathematics
that are beyond the scope of this study—however, there are many applications available
to make creating cartograms relatively simple. In the case of this study, a contiguous
areal cartogram was created using an ArcMap cartogram geoprocessing add-on, which
was created by Tom Gross of Esri. This tool was obtained from the ArcScripts repository.
Contiguous area cartograms have the benefit of dramatically emphasizing the important
data—the actual number of Electoral votes won by each candidate—which can eliminate
the confusion that often results from more traditional election mapping techniques.
Having illustrated the most common mapping techniques, we now turn to an
exploration of how these different methods of representing election outcomes on maps

30

can enhance or detract from our interpretation of election results with respect to the
geographic distribution of votes and population.

31

CHAPTER 5: MAPPING THE ELECTION RESULTS
While determining the winner of an election based on any of the proposed reform
methods is a relatively simple process, representing the election results in a way that is
informative and is not misleading is anything but. The United States is a vast country,
sparsely populated in some areas (Wyoming had a population of just over 582,600 at the
2010 census) and densely populated in others (California being the most populous state,
with over 38,332,500 citizens at the 2010 census). The areal size disparity of the states
poses another challenge: The smallest state, Rhode Island, has a total size of only 1,544
square miles—while the largest state, Alaska, is a massive 665,384 square miles (Johnson
2009). These differences make representing election results in a way that is easily
interpreted by the viewer challenging. How does one present the information that is
important, while still preserving the integrity of the geography? Does the inherent bias of
all maps make this an impossible task? If so, what is the best compromise?
Introducing the different reform methods only adds to the challenge, because they
increase the amount of information that needs to be represented. And, because the maps
of reform methods present concepts that the general public isn’t familiar with, it is all the
more important that the information is presented in a manner that is both clear and
accurate.
Given the different geographies that are used in the various election methods, the
kind of cartographic technique that best represents each method differs. These
differences are explored in the following sections. Table 6 provides an overview of the
exploration.

32

Table 6: Comparison of techniques for mapping election results

Reform method
Map Styles Current
Proportional
Allocation
Congressional
Districts
Popular
Vote
Simple Red/Blue map ü X üü X
Choropleth map üü üü X X
Proportional Symbols ü üü X X
Dot density X X X üü
Cartogram ü X X X
One check indicates the technique can be used to represent that method, two checks indicate the
best method. X indicates the method is not appropriate.

5.1 Mapping the Traditional Electoral College Method
As mentioned earlier, the areal differences between the boundaries of America
can create visual confusion when attempting to represent election results that are awarded
per state—the smaller states are often overlooked despite the large number of Electoral
votes they might carry, while the larger states (which often carry few Electoral votes due
to low population) give the impression that they have a bigger impact on election
outcomes than they actually do.
As indicated in Table Six, one approach to more accurately represent the election
outcome is the use of choropleth mapping techniques. By assigning a bolder color to the
states that have a higher number of Electoral votes and a lighter color to the states that
have fewer votes, it is possible to create a map that gives the viewer a better
understanding of the true outcome of the election. The map is still dominated by red
tones, but it is much more clear to the viewer that the majority of the large states that
went red don’t carry a large number of Electoral votes (as illustrated by their pale
coloring), while the smaller states on the East coast actually contribute a significant
number of Electoral votes (as illustrated by their darker blue coloring) (Figure 4).

33

Figure 4: Electoral College results for the 2012 Presidential Election, presented in a
Choropleth format based on the amount of electoral votes awarded by each state

Choropleth mapping is simple and effective, but does leave out some data that
some viewers might find important to the story of the election outcome, such as
population. Bringing in more information can help the viewer to understand that the
areal size of the state does not necessarily correlate with either population or electoral
votes awarded. One example of this is to pair a choropleth representation of the
population of each state rather than the Electoral votes awarded. This approach de-
emphasizes the visual impact of the red/blue color scheme, but still conveys the Electoral
information with graduated symbols that illustrate the power that the individual states
carry and which candidate won each state (Figure 5).

34

Figure 5: Electoral College results for the 2012 Presidential Election, presented in a
combination choropleth and graduated symbol format based on population and the
amount of electoral votes awarded by each state.

This map format is the most informative thus far, because it better conveys both
the Electoral impact of each state, and why the number of Electoral votes per state varies
so much throughout the country. Unfortunately, the map still suffers from the size
disparity amongst the states—the smaller East Coast states all but disappear when placed
next to the larger states in the West, but they are in fact home to a sizeable portion of the
population. Additionally, it is a difficult map to interpret, because the viewer must
understand the meaning of two very different symbologies, and combine them together to
understand what the map is intended to relay.

35

One way to get around the consistent areal problem is to create a cartogram
(Figure 6). As previously explained, cartograms are maps that are designed to sacrifice
geographic integrity in order to better represent the impact of a particular variable. In the
case of an Electoral College map, this variable would be the number of Electoral votes
carried by each state.

Figure 6: Electoral College results for the 2012 Presidential Election, presented as a
cartogram where area is proportional to the number of electoral votes
Note: Insets shown at the same scale as main map

Cartograms solve the problem of a visually misleading map in one way—it is now
much more clear that the Democratic candidate was the winner of the election—but they
create a new problem, in that they render the country almost unrecognizable. The
outlines of the geography of the United States can be added as an overlay (Figure 7),

36

which is helpful—but this adds to the visual noise of the map and distracts from the
results that the Cartogram is representing.

Figure 7: Electoral College results for the 2012 Presidential Election, presented as a
cartogram with the outlines of the United States as an overlay to demonstrate the extent
to which the geography has been altered.

The distortion of a cartogram might not be an issue when the information is being
presented to an American audience who would have at least a passing understanding of
the geography of their home country. To an audience less familiar with the geography of
the country, however, the distortion created by the cartogram adds a degree of confusion.
In short, if maintaining the geographic integrity of the country is important to the story
that the map is telling, a cartogram is not a viable option.

37

5.2 Mapping the Popular Vote Method
The Popular Vote method presents a unique challenge when it comes to creating a
results map, because the results are not actually tied to geography in any way. Of course,
the citizens who are doing the voting have a geographic presence—they live in a
particular city, county, and state—but a geographic unit does not alter the impact of their
vote in the same way that it would with the other proposed reform methods. In spite of
this, and because the geography in which we live can have a strong correlation with the
way we think (and, therefore, the way we vote), choosing the appropriate representation
of the voting outcome is important. Looking back at Table 6, it quickly becomes clear
that almost all of the mapping techniques are inappropriate, because they are all
dependent on data that is tied to geography in a strong way.
Despite the fact that the popular vote method is not tied to geography, it is
nonetheless useful to create a map that somehow represents the individuals in their spatial
context. The scale of the country presents a problem (as discussed previously), but so
does the large population. How can each of the 122,188,009 votes of the 2012 election
be represented on a map in a clear and concise way? It would perhaps be possible if there
were no attention paid to geography whatsoever, however the strong correlation between
population density and voting behavior means that the location of the voter cannot be
entirely ignored (De Chant 2013).
Because of this correlation between population density and voting, a dot density
map is the most useful approach to representing the popular vote results of the
presidential election. Representing the results of the Popular Vote method nationwide,
though accurate, does not provide a clear enough picture of how and where people are

38

voting—therefore some aggregation is required to give a better overall picture of the
election. In dot density mapping, dots are dispersed randomly within the boundaries of
the area they represent, so rather than showing results at state level, these dots were
grouped at the county level, in order to give a better understanding of where the majority
of Americans live and vote (Figure 8). Although the election data is initially collected at
the precinct level, the detail of the results would not be visible at such a small level of
aggregation when viewed at the national scale.
While the resulting map doesn’t clearly convey who won the election, the dense
clustering better indicates the voting power that these tiny states carry, while the sparse
areas tell the viewer that, though large in size, those states do not contain a large voting
population. Increasing the size of the dots and the number of votes that each dot
represents helps matters somewhat, but the representation still suffers from over-
crowding (Figure 9), which obscures much of the data as the results for one party overlap
the other. The overlap problem can be mitigated by producing a results map for each
party (Figures 10 and 11), but it creates a new problem, because it is difficult to compare
the results without them falling on the same map.

39

Figure 8: Voting results of the 2012 election nationwide, in which a single point
represents 10,000 votes and dots are constrained by county.

40

Figure 9: Voting results of the 2012 election nationwide, in which a single point
represents 50,000 votes and dots are constrained by county.

41

Figure 10: Republican Party voting results of the 2012 election nationwide, in which a
single point represents 50,000 votes and dots are constrained by county.

42

Figure 11: Democratic Party voting results of the 2012 election nationwide, in which a
single point represents 50,000 votes and dots are constrained by county.

All of the maps are somewhat difficult to interpret because the size of the country
and the sheer number of votes being represented obscures any detail that might otherwise
have been discernible and because the necessary symbology makes it impossible to know
how votes were cast in counties with fewer than 50,000 votes. However, they do allow
the viewer to draw an informed conclusion about both the winner of the majority of the
votes and the areas of the United States that have the strongest voting numbers.
Now that the election outcomes have been examined as both a table and a map, it
becomes a little easier to address the claims of the proponents of the popular vote method,
as well as the concerns of some of its detractors. This method without a doubt represents
the choices of the voters, just as the proponents claimed. Because there are no electoral

43

votes coming into play, the vote totals are an exact representation of what the voters
chose. The question of whether or not the Popular Vote method would eliminate swing
states is more difficult to answer. Clearly, there are no states involved in the calculation
of the election totals, but at the same time, there are still some regions of the country in
which the voters are more likely to be evenly divided and/or undecided—and those
regions might attract the attention of campaigning presidential hopefuls in much the same
way that swing states historically have, which could create “swing regions” rather than
eliminate the swing state problem as many proponents hope.
If swing states were in fact to disappear, however, it is entirely possible that
elections could become more expensive, as opponents of this method assert. If
candidates were suddenly required to campaign across the entire country with the same
focus that they currently reserve for swing states, campaign costs could skyrocket.
Alternatively, costs could remain the same and campaign strategies could change focus to
create a more moderate level of campaigning for the entire country.
Finally, it is evident in looking at the maps that the Popular Vote method would
weaken the power of some of the less populous states—because they have fewer
individual votes to offer. In fact, this method weakens the power of every state and
increases the power of the individual, because states have no bearing in the calculation of
the election winner. Whether or not this is a weakness of this method or a benefit is
entirely a matter of perspective.

44

5.3 Mapping the Proportional Allocation Method
The Proportional Allocation Method is an attempt to bridge the gap between the
traditional Electoral College and the Popular Vote Method. Electoral votes are still
awarded, which negates the fears of states losing their power—but they are awarded
based on the outcome of the popular vote, which addresses the concerns of those who feel
that the voters are under-represented with the traditional method.
Mapping the results using this method is a matter of representing proportions of
total votes per state, which as indicated in Table 6 is a good fit for the proportional
symbol mapping technique. Unfortunately, while the proportional symbol technique is
effective in demonstrating which areas have large differences in vote distribution, it falls
short in areas where the vote distribution is similar, but not exactly the same. For instance,
in Figure 12 (below), it is clear that Obama won in California (where he had an eleven
point lead), because there is a noticeable difference in size between the two circles. It is
much more difficult to determine the winner in Florida, however (where Obama had only
a one point lead), because the two circles are nearly identical in size. Beyond that, the
circles make it almost impossible to tell exactly how many votes each candidate received.
Looking at New York, it is evident that Obama received a higher number of votes, but
how many electoral votes was he awarded? How many went to Romney? The circles
don’t look dramatically different in size, but in reality Obama actually won almost twice
as many electoral votes (18 to Romney’s 11).

45

Figure 12: Proportional symbol map of the 2012 election, using the proportional
allocation method

This difficulty on the part of map viewers to determine small variations in the size
of different circles and make accurate comparisons between them has been studied, and
some solutions have been offered to make the data representation easier to understand.
Esri offers an “appearance compensation conversion” that alters the size of the circles
used to make it easier for viewers to interpret the difference in circle sizes, while others
feel that using graduated symbols with a set range of values (Figure 13) and a legend
showing a variety of circle sizes is preferable (Meihoefer 1973).

46

Figure 13: Graduated symbol map of the 2012 election, using the proportional allocation
method

For a more precise representation of the voting distribution, proportional symbols
are abandoned in favor of bar graphs (Figure 14). The bar graphs function in much the
same way, but they do away with the likelihood of misinterpretation of the proportional
circles. This results in a map that displays the data more accurately and with a more
discernable difference between two similar but unequal values (again, such as in Florida)
although it is still hard to interpret exact values in states where there are a very small
number of total Electoral votes (such as in Wyoming, which has a grand total of three
votes to hand out).

47

Figure 14: Bar chart map of the 2012 election, using the Proportional Allocation method
and showing total population per state

Proponents of Proportional Allocation argue that the method will create a more
“National” election, because the lack of a guarantee of a set number of electoral votes
would eliminate the swing states. As was previously discussed, however, there is no real
way of knowing if this would be true. The likelihood of swing states would be reduced,
certainly, but there are still many states within the country with an even or almost even
vote distribution in the 2012 election—it’s likely that such states would be the focus of
future campaigns under this method. Opponents of Proportional Allocation again argue
that it could weaken the power of the less populous states, but the number of votes

48

awarded per state remains the same (until the next Census)—they are merely distributed
differently.

5.4 Mapping the Congressional District Allocation Method
Mapping election results using the Congressional District method comes with its
own unique set of challenges, which, as indicated in Table 6, limits the choices of
appropriate map techniques available. Choropleth mapping, cartograms, and proportional
symbols are not an option for this method, as each District awards the same number of
votes: one. At first glance, it seems as though the same standard red/blue representation
methods that were used for the traditional Electoral College results can be applied, since
the districts are won on a winner-take-all basis (Figure 15). This is true, with one major
difference: the extra two Senatorial votes awarded to the winner of each state. Those two
extra votes are challenging to represent clearly on a map, because they are awarded using
a different geography than the bulk of the data.

49

Figure 15: House of Representatives district results for the 2012 Presidential election,
excluding the two Senatorial votes per state

In order to represent the voting results more accurately, a transparent state-level
overlay can be created, showing the color of the overall winner of each state (and thus the
winner of the two Senatorial votes). This adds a slight degree of confusion, as the
transparent overlay will turn the opposing colors purple, but it ensures that important
information about the winner of the majority of the votes isn’t lost (Figure 16).

50

Figure 16: Congressional District results for the 2012 election, with state-level overlay

As with the traditional Electoral College maps, there is still an issue created by
the vastness of some of the states, making the outcome of the election difficult to
determine at a glance. Unfortunately, cartograms are not an option in this instance, for
two reasons: One, each Congressional District is only worth one Electoral vote, so the
cartogram cannot be calculated on that field as it was previously; two, Congressional
Districts are determined by population. There are a fixed number of 435 districts total in
the United States, each one of which is supposed to carry a relatively equal number of
citizens. After each decennial census, the boundaries of the districts are redrawn to
reflect the changes in population per state and maintain an equal population distribution.
Currently, each Congressional District is home to an average of 710,767 people—so,

51

creating a cartogram based on population wouldn’t work either, because each district
carries a relatively equal weight.
Now that the Congressional District method has been examined in both table and
map form, it is possible to gain a clearer perspective on the arguments for and against this
type of Electoral College reform. Proponents of this method claim that it better
represents the popular vote, however the results tables for both the 2000 and 2012
elections clearly shows that in both cases, the winner of the popular vote would not have
been the winner of the election if the Congressional District method had been used. It is
true that this method has been used in Maine and Nebraska without incident, which is
something that none of the other proposed reform methods can claim. When it is applied
at a national level, however, the effects of a changing geography (whether through
Gerrymandering or simple redistricting without political motivation) lead some to
question whether the Congressional District method would be an effective change from
the Electoral College.
Having explored each of these methods from the perspective of both tables and
maps, what have we learned? What has the exploration of mapping techniques told us
about not only the reform methods and their relationship to the choices of the voting
public, but about the challenges of mapping election results in general? The final chapter
reviews the research questions posed in Chapter 1, determining what conclusions can be
reached and what topics require further exploration.

52

CHAPTER 6: CONCLUSIONS
Many people might believe that geography has absolutely nothing to do with Presidential
elections, beyond the fact that you must be a citizen of the United States in order to vote.
Without an understanding of how the Electoral College functions—and how those
methods that are being offered as a superior replacement function—it might seem as
though the total number of votes per candidate is the only thing that matters. Upon
further exploration, however, the truth becomes evident: geography has the power to
change elections. In some cases, this simply refers to the fact that the geographic
location of the voters can impact their political choices (De Chant 2013), while in other
cases this means that geographic boundaries can actually be manipulated in order to be
more likely to produce the outcome being sought by a certain political party. If it is
resolved that the Electoral College is indeed in need of reform, it is vital that those who
are the architects of said reform pay close attention to the manner in which votes are
calculated, so that they can ideally choose a method that aligns most closely with the
wishes of the voting public—even when the effects of geography are taken into account.
This study explored the Electoral College and the geography of elections in two
very different ways: in the first, tables were created in order to provide a quick and clear
answer to the question of whether the proposed reform methods would change election
outcomes. Once it was determined that they would, in fact, change the outcomes, the
study then turned to the question of why they would change (and whether those changes
validated the claims of each method’s proponents) by undertaking an in-depth visual
analysis of each proposed reform method via a variety of maps created using GIS
software. In doing so, it provides the reader with multiple resources, not only for

53

understanding how each proposed reform method functions, but also for understanding
the intricacies of election mapping in general.
This investigation is not going to end the debate about how best to reform the
Electoral College—or even the debate over if it needs to be reformed, at all. It is a
complicated issue that has been argued about by political scientists for years, and will
most likely be argued about for years to come. However, one can now see that of the
many pros and cons offered for each potential reform method, one that is seldom
mentioned—geography—may in fact have the biggest impact. In the case of the hotly
contested 2000 Presidential election, for example, two of the proposed reform methods
would have resulted in the election of Al Gore as President, while the third would have
elected George Bush (as the traditional Electoral College system did). If accurately
representing the choices of the voters is one of the driving forces behind the reform
proposals, then close attention must be paid to the fact that only two of the proposed
reform methods accomplish that task consistently: Popular Vote and Proportional
Allocation. Of course, our founding fathers might argue that representing the choices of
the voters was never a priority in the creation of the Electoral College.
The question of how best to represent both the results of an election
geographically and the location of the voting public also does not have one definitive
answer. The vast and varied geography of the United States will always pose a unique
challenge in attempting to represent election outcomes, and there is no one map to suit
every audience. Instead, at the onset, the map creator must ask him/herself what
information they wish to convey to whom, and what map elements will enhance or
detract from the viewer’s understanding of that information. The map viewer, in turn,

54

must bear in mind the challenges a map maker faces (inconsistent geography, data
limitations, and personal biases, among other things) and use that knowledge to
understand that no one map can tell the whole story—but every well-constructed map can
still tell the parts of the story that are most important.
There is, of course, still much to be explored on the topic. Future studies might
include an in-depth exploration of the current campaign process (in which candidates
focus the bulk of their money and energy on swing states), and the question of whether
changing the vote calculation process can eliminate swing states altogether, or whether
they would simply morph into swing counties/districts. This topic was briefly discussed
in Chapter 5, but an in-depth analysis of how swing states are created and their effect on
campaign strategy would no doubt yield valuable insight into the likelihood of new swing
states/districts in the event of Electoral College reform.
Additionally—or perhaps even concurrently—there are still many avenues to
explore in the realm of election mapping. While this study encompassed some of the
most commonly found symbology methods, the availability of newer technology such as
three-dimensional mapping software creates more possibilities for election mapping.
What information can the added dimension convey that a traditional two-dimensional
map cannot? How can this expand on a map viewer’s understanding of the complexities
of voting behavior and geography on election results? As technology progresses, this
topic will no doubt continue to expand, creating more and more possibilities for the
mapping of voters and election outcomes.

55

REFERENCES

Anderson, J.B., 2001. Flunk the Electoral College, Pass Instant Runoffs. The Progressive, 65(1),
pp. 17-19.

Belenky, A., 2008. The Good, the Bad and the Ugly: Three Proposals to Introduce the
Nationwide Popular Vote in U.S. Presidential Elections. Michigan Law Review First
Impressions, 106(4), pp. 110.

Belenky, A., 2013. Who Will Be the Next President? A Guide to the US Presidential Election
System. Springer ebook. http://www.springer.com/law/book/978-3-642-32635-6 (last
accessed 21 April 2014).

Black, E., 2012, 10 Reasons Why the Electoral College is a Problem.
http://www.minnpost.com/eric-black-ink/2012/10/10-reasons-why-electoral-college-
problem (last accessed 15 April 2013).

Deschamps, D., Farrell, L.D. and Singer, B., 2012. Electoral Dysfunction. Movie. New York:
Trio Pictures.

De Chant, T., 2012. How Population Density Affected the 2012 Presidential Election.
http://persquaremile.com/2012/11/08/population-density-and-the-2012-presidential-
election/ (last accessed 25 March 2013).

Enda, J., 2012. When Republicans Were Blue and Democrats Were Red.
http://www.smithsonianmag.com/history/when-republicans-were-blue-and-democrats-
were-red-104176297/?no-ist (last accessed May 2014).

Gelman, A., 2014. The Twentieth-Century Reversal: How Did the Republican States Switch to
the Democrats and Vice Versa? Statistics and Public Policy, 1(1), pp. 1.

Hirsch, S., 2008. Awarding Presidential Electors by Congressional District: Wrong for
California, Wrong for the Nation. Michigan Law Review First Impressions, 106(4), pp.
95.

Jenkins, J.A. and Sala, B.R., 1998. The Spatial Theory of Voting and the Presidential Election of
1824. American Journal of Political Science, 42(4), pp. 1157.

Johnson, J., 2009. United States Electoral College. New World Encyclopedia.
http://www.newworldencyclopedia.org/p/index.php?title=United_States_Electoral_Colle
ge&oldid=926147 (last accessed 3 April 2013).

Kimberling, W.C., 1992. The Electoral College. http://www.fec.gov/pdf/eleccoll.pdf (last
accessed 24 April 2013).

56

Koehler, R., 2010. Gerrymandering. Variety, 420(8), pp. 52.

Levy, M., 2012. The Electoral College: Outdated Artifact of History. Encyclopedia Britannica.
http://www.britannica.com/blogs/2012/10/electoral-college-outdated-artifact-history/
(last accessed 10 October 2013).

Longley, L.D. and Peirce, N.R., 1996. The Electoral College Primer. New Haven, CT: Yale
University.

Meihoefer, H.J., 1973. The Utility of the Circle as an Effective Cartographic Symbol. The
Canadian Cartographer, 10(1), pp. 63

Monmonier, M, 1991. How To Lie With Maps. Chicago, IL: University of Chicago Press.

Neale, T.H., 2009. Electoral College Reform: 111th Congress Proposals and Other Current
Developments. R40895. Washington: Congressional Research Service .

Neale, T.H. and Whitaker, L.P., 2004. The Electoral College: An Overview and Analysis of
Reform Proposals . RL30804. Washington: Congressional Research Service.

Oxford Dictionaries, s.v. Swing States. http://oxforddictionaries.com/ (last accessed 13 April
2013).

Rose, G. L., 1994. Controversial Issues in Presidential Selection. Albany, NY: SUNY Press.

Slocum, T.R., McMaster, R.B., Kessler, F.C., and Howard, H.H., 2009. Thematic Cartography
and Visualization (Third Ed.). Upper Saddle River, NJ: Pearson Prentice Hall.

The Center for Voting and Democracy, 2009. Reform Options for the Electoral College.
http://archive.fairvote.org/e_college/reform.htm (last accessed 25 March 2013).

The Center for Voting and Democracy, 2009. Maine & Nebraska.
http://www.fairvote.org/reforms/national-popular-vote/the-electoral-college/solutions-
and-the-case-for-reform/maine-nebraska/ (last accessed 1 May 2014).

U.S. Constitution, Article II, Section I; Amendement XII.

U.S. Senate. 1970. Hearings Before the Committee on the Judiciary United States Senate: 91st
Congress, 2nd Session, On Amending the Constitution Relating to Electoral College
Reform, 15-17 April 1970, U.S. Senate Committee on the Judiciary. Washington:
Government Printing Office.

Vile, J. R., 2005. The Constitutional Convention of 1787: A Comprehensive Encyclopedia of
America's Founding. Santa Barbara, CA: ABC-CLIO, Inc.

Abstract (if available)

Abstract Multiple events throughout the history of the United States of America have led people to call for the Electoral College system to be reformed or abandoned altogether. As the Electoral College currently functions, each state awards a set number of votes (determined by population) to the candidate who receives the largest number of votes, but many citizens feel that there are flaws in this system. Although there have been many reform propositions over the years, there are three potential methods that consistently have the most support: Popular Vote, Proportional Allocation, and Congressional Districts Allocation. This study offers insight into how each of these reform methods might change election outcomes and even more importantly, by exploring several possible election mapping techniques, it provides an analysis of how the presentation of election results in a geographic format can alter the viewer’s perceptions of election outcomes and of the viability of the various reform methods. Finally, this study provides arguments for why the traditional methods of representing election outcomes tend to fall short.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The geography of voter power in the U.S. electoral college from 1900-2012

PDF

Drawing better lines: comparing commissions to legislatures on compactness and coterminosity

PDF

Mapping uniformity of park access using cadastral data within Network Analyst in Wake County, NC

PDF

Wake County District Overlay: an online electoral data visualization application

PDF

Social media to locate urban displacement: assessing the risk of displacement using volunteered geographic information in the city of Los Angeles

PDF

Water Rights Permit System (WRPS): a GIS‐based tool for the Umpqua Drainage Basin

PDF

Spatiotemporal visualization and analysis as a policy support tool: a case study of the economic geography of tobacco farming in the Philippines

PDF

A fire insurance map geocoder for pre-earthquake San Francisco

PDF

Determining the utility of GIS in border disputes case study: Sudan and South Sudan

PDF

Exploring urban change using historical maps: the industrialization of Long Island City (LIC), New York

PDF

Is the likelihood of waterfowl presence greater on conserved lands?

PDF

Improving the communication of water allocation decisions using interactive maps

PDF

Evaluating transit and driving disaggregated commutes through GTFS in ArcGIS

PDF

A comparison of GLM, GAM, and GWR modeling of fish distribution and abundance in Lake Ontario

PDF

Integrating spatial visualization to improve public health understanding and communication

PDF

Finding food deserts: a study of food access measures in the Phoenix-Mesa urban area

PDF

Mapping firing ranges as social capital generators in Houston, Texas

PDF

Effect of spatial patterns on sampling design performance in a vegetation map accuracy assessment

PDF

An evaluation of soil sampling methods in support of precision agriculture in Northeastern North Carolina

PDF

Testing LANDIS-II to stochastically model spatially abstract vegetation trends in the contiguous United States

Asset Metadata

Creator Calahan, Meagan Eileen (author)

Core Title Investigating electoral college reform: geography's impact on elections, and how maps influence our perception of election outcomes

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 07/07/2014

Defense Date 05/23/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag electoral college,GIS,mapping techniques,OAI-PMH Harvest,presidential elections

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Kemp, Karen K. (committee chair), Vos, Robert O. (committee member), Warshawsky, Daniel N. (committee member)

Creator Email calahan@usc.edu,meaganvm@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-431787

Unique identifier UC11286587

Identifier etd-CalahanMea-2627.pdf (filename),usctheses-c3-431787 (legacy record id)

Legacy Identifier etd-CalahanMea-2627.pdf

Dmrecord 431787

Document Type Thesis

Format application/pdf (imt)

Rights Calahan, Meagan Eileen

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA