Understanding Presidential Voting Motivation by Factors of Agency

Chapter 1: Introduction to the Study

Background of the Study

The selection of executive leadership at the highest level of

government that accurately reflects the values and traditions of the people

the government represents requires that the represented populace go to the

polls and cast a vote. Voting has traditionally been considered an act of

agency: an exercise of volitional behavior (Cwalina, Falkowski, & Newman,

2008). Voting agency theory offers are three broad answers to the question of

who (or what) exactly exercises volition. Personal agency theory suggests

that the individual makes unmediated voting decisions that are influenced,

but not determined, by other people or environmental circumstances in the

voter’s life. Social agency theory suggests that the individual’s choice is

largely determined by the people they have intimate relationships with

(family, friends, neighbors, influencers etc.). Finally, sociocultural agency

suggests that it is neither the unmediated individual nor a small social

grouping of like-minded others that is responsible for exercising voting

choice. According to this body of theory, voting choices emerge from the

actions of entire social classes.

These three research traditions in voting agency theory have

historically been largely distinct from each other. Sociocultural theorists,

starting with Marx (1972) and more recently including scholars such as

Thomas Frank (2004) and Elizabeth Beaumont (2011) have tended to

perform qualitative work focusing on the influence exercised by large groups

of people (such as the working class) on individuals’ political behavior; such

work has extrapolated from the behavior of smaller groups to larger groups.

Personal agency theorists often work in the neoclassical and individualist

tradition of Adam Smith (1801); this tradition is conceptually, theoretically,

and methodologically irreconcilable with many of the assumptions of

sociocultural theorists. Social agency theorists, such as Robert Putnam

(2001a, 2001b), have a distinct research program influenced by sociocultural

theory and personal agency theory but that has developed its own vocabulary

and concerns.

Recent empirical research on the topic of agency theory in the scope

of political science has tended to be limited to either sociocultural agency,

personal agency, or social agency. For example, Lin (2010) focused on

voters’ personal attitudes to aspects of governance, Poundstone (2008)

examined the role of families and other social units as determinants of

individual voting preferences, and Campbell (2006) examined the predictive

power of voting. The applicability of these study findings, however, is

generally limited by a low coefficient of determination. Scholars have tended

to conduct empirical research bounded by specific agency theories, instead

of taking a broader approach that encompasses multiple agency theories. As

a result, their regression models lack explanatory power (Aggarwal,

Meschke & Wang, 2012; Gonzalez-Ocantos, De Jonge, Melendez, Osorio &

Nickerson, 2012; Kasfir, 2014).

This study was designed to address these limitations by incorporating

multiple forms of agency into its empirical model. Kiewiet and Lewis-Beck

(2011) suggested that voting is an extraordinarily complex act and is likely

influenced by a multitude of personal and social factors. In order to test the

importance of these factors, and to determine which kinds of agency might

be more influential in predicting voting, it is necessary to conduct a form of

regression in which the independent variables include several kinds of

agency variables. The remainder of the chapter consists of the problem

statement, purpose of the study, research questions and hypotheses,

theoretical framework, nature of the study, definitions, assumptions, scope

and delimitations, limitations, and significance.

Problem Statement

Voter turnout in the United States remains lower than in many nations

(Simon, Egidi, & Viale, 2008). This low turnout erodes the quality of

democracy as only sections of the population are significantly represented

(Postman, 2006). In order to raise voter turnout, it is necessary to learn more

about why people vote, so that policy actions can be better targeted to

improving participatory behavior (Putnam, 2001a, 2001b). Although

scholars have enumerated the general reasons that people vote, these reasons

have not been weighed against each other (Benney, Gray & Pear, 2013; van

der Straeten, Laslier & Blais, 2013). An example of this is the proposition by

Sexton (2011) that for people to vote the benefits need to outweigh the cost.

The determinants of voting behavior are difficult to compare precisely

because of the radical subjectivity of voters’ cost-benefit analyses (Sosnaud,

Brady & Frenk, 2013; Stevenson & Wolfers, 2013).

There is no preexisting quantitative framework through which

scholars can understand the relative contributions of personal, social, and

sociocultural agency to voting behavior. Although there is a scholarly

consensus that there are many independent variables that predict voting

(Sexton, 2011), there is a fecund amount of study which has sought to

examine this problem through the utilization of statistical models. Academic

study is therefore needed in order to inform policymakers and scholars who

the people are that will participate in voting as indicated by the significance

of the strength of agency factors as influencers of the behavior of voting.

Purpose of the Study

The primary purpose of this quantitative study is to explore the

sociotropics of voting behavior. This exploration is performed through

determining how voting emerge from a combination of personal, social, and

sociocultural agency (Kiewiet & Lewis-Beck, 2011). The statistical model of

this study measures personal agency through subject perceptions of four

variables: (a) how much people think that they can affect government, (b)

the importance that respondents attach to people doing what authorities tell

them, (c) whether or not respondents are optimistic or pessimistic about their

nation’s future, and (d) respondents’ interest in information about

government and politics. Social agency is measured through the degree of

respondent’s family’s interest in politics. Socoiocultural agency is measured

through the following variables: (a) household income, (b) home ownership,

and (c) education. The goal of the related statistical analysis was to quantify

the predictive power of each of these independent variables on the dependent

variable of election vote, and in doing so to arrive at a conclusion about

which of the three varieties of agency theory is more predictive of voting.

Research Question and Hypotheses

The study was examined American voting by analyzing data collected

from the

American National Election Studies (ANES) series of surveys. These

surveys were administered to American voters in every presidential election

year since 1948.The primary research question for this study was:

RQ: How did the factors of personal, social, and sociocultural agency

influence

American voters’ stated voting actions in U.S. Presidential elections

from 1952 to 2008?

The null hypothesis for the primary research question was that neither

personal, social, or sociocultural agency factors had any predictive power in

determining whether a representative sample of American voters intended to

vote in U.S. Presidential elections. The alternative hypothesis was that at

least one factor was a significant predictor of American voting in U.S.

Presidential elections. Twenty specific subhypotheses were developed to test

specific factors:

H10: There is no significant relationship between age and voting in U.S.

Presidential elections.

H1A: There is a significant relationship between age and voting in

U.S.

Presidential elections.

H20: There is no significant relationship between gender and voting in

U.S.

Presidential elections.

H2A: There is a significant relationship between gender and voting in

U.S.

Presidential elections.

H30: There is no significant relationship between race and voting in

U.S.

Presidential elections.

H3A: There is a significant relationship between race and voting in

U.S.

Presidential elections.

H40: There is no significant relationship between education and voting

in U.S.

Presidential elections.

H4A: There is a significant relationship between education and voting

in U.S.

Presidential elections.

H50: There is no significant relationship between family income and

voting in U.S. Presidential elections.

H5A: There is a significant relationship between family income and

voting in U.S.

Presidential elections.

H60: There is no significant relationship between religion and voting

in U.S.

Presidential elections.

H6A: There is a significant relationship between religion and voting in

U.S.

Presidential elections.

H70: There is no significant relationship between home ownership and

voting in U.S. Presidential elections.

H7A: There is a significant relationship between home ownership and

voting in U.S. Presidential elections.

H80: There is no significant relationship between interest in the

elections and voting in U.S. Presidential elections.

H8A: There is a significant relationship between interest in the

elections and voting in U.S. Presidential elections.

H90: There is no significant relationship between interest in public

affairs and voting in U.S. Presidential elections.

H9A: There is a significant relationship between interest in public

affairs and voting in U.S. Presidential elections.

H100: There is no significant relationship between a belief that one has

any say in what government does and voting in U.S. Presidential elections.

H10A: There is a significant relationship between a belief that one has

any say in what government does and voting in U.S. Presidential elections.

The research question was designed to be addressed through the use of

logistic regression, in which the null hypothesis is accepted if the p value of

each individual agency factor’s contribution to voting is ≥ .05, and the null

hypothesis is rejected if p < .05 for any of the agency variables. The

inclusion of demographic as well as agency variables allows the answer to

the research question to reflect variation due to different facets of

respondents’ circumstances and identities. The ANES dataset from which the

personal, social, and sociocultural agency variables were taken, and the

nature of these variables themselves, is discussed further in chapter three.

Theoretical Base

One way of organizing a discussion of the theoretical base is through

the notion of agency. Agency theory has three main divisions: (a) personal

agency, in which theorists assume that humans are want-maximizing,

purposive decision-makers who consciously seek to achieve goals that

gratify themselves; (b) social agency, in which theorists assume that social

units (such as families and neighborhoods) help to determine individual

decisions, based on their influence of the individual; and (c) sociocultural

agency, in which theorists assume that individual decisions are the products

of membership in large groups (for example, the working class or the

Republican Party).

There is incidental amounts of overlap between the various kinds of

agency theory. For this reason, agency theory is better described as being a

spectrum than a dichotomy (Cote & Levine, 2014).For example, personal

agency theorists acknowledge that decisions do not take place at a perfect

Archimedean remove from the structural and infrastructural facts of an

agent’s life, many of which are predetermined before the agent’s birth

(Rogers & Shoemaker, 1971). Yet, like classical liberals, personal agency

theorists emphasize that voter behavior is essentially want-maximizing

behavior in response to felt needs and based on an ultimately rational

appraisal of their situations (O’Sullivan, 2009). Sociocultural theorists (e.g.,

Thoms, 2006) acknowledge some modicum of personal choice and want

maximization. However, they also focus more acutely on the background

context of this behavior than social theorists of agency, showing it as either

wholly (e.g., Marx, 1972) or partly determined by variables that have little,

if anything, to do with individual choice.

There is a degree of theoretical complexity which exists in the current

study as there are social and cultural factors under examination. While social

and cultural complexities exist, quantitative study can still have implications

for theory if the limitations present are acknowledged. As an example, if

quantitative identifies that making over $175,000 a year is the surest

predictor of voting, then rationale becomes evident that there is a need for

further investigation of a general sociocultural theory of voting built around

income. These results cannot be presented as either simple or causal, but

must be taken as planks in an ongoing debate about voter motivations that

neither this study nor any other is likely to conclusively resolve.

Nature of the Study

This quantitative study used logistic regression as its main statistical

technique for data analysis. The study’s dependent variable was voting in

U.S. Presidential elections.

Its independent variables of the study were:

1. potential voters’ perceptions of how much people can affect

government;

2. potential voters’ perceptions of how important it is to do what

authorities tell

them;

3. potential voters’ optimism or pessimism about the nation’s future;

4. potential voters’ interest in information about government and

politics;

5. potential voters’ families’ interest in politics;

6. potential voters’ household income;

7. potential voters’ home ownership status; and

8. potential voters’ education.

The intermediate variables of the study were race, gender, major party

affiliation, religion, ethnicity, and age. The data pertaining to these variables

was collected from the published ANES dataset pertaining to U.S.

Presidential elections. This dataset contains contains responses from

hundreds of thousands of potential voters who were statistically

representative of the American electorate in various election years.

Definition of Terms

Agent. An individual in his or her capacity as a decision-maker,

especially when such decisions take place in relative conditions of autonomy

(Krugman & Wells, 2009).

Agentic. Behavior that reflects the relative autonomy of an agent

(Krugman &

Wells, 2009).

Mobilization. Pro-voting messages (such as through the use of

advertising, both public and private) that reach the voter and influence his or

her behavior (Putnam 2001a,

2001b).

Personal agency theory. Theory rooted in the belief that humans are

wantmaximizing, purpose actors who seek to achieve certain ends for their

own individual reasons (Smith, 1801).

Self-interest. A state of being in which an agent attempts to reach his

or her goals, regardless of the implications of this want-maximizing behavior

for others

(Poundstone, 2008).

Social agency theory. A theory stating that smaller social units (such

as families, neighborhoods, business cultures, etc.) function as determinants

of individual political behavior (Poundstone, 2008).

Sociocultural agency theory. As articulated by theorists including

Frank (2004) and Marx (1974), a theory claiming that larger social and

economic units such as economic classes and political parties) play the

primary role in determining individuals’ economic behavior.

Sociotropic. A voting behavior that is understood to reflect some

combination of personal, social, and sociocultural agency (Kiewiet & Lewis-

Beck, 2011).

Uses and gratifications theory. A personal agency theory committed

to the position that voters vote in order to satisfy personal gratifications,

such as the need to participate in a democracy (Rubin, 1983).

Assumptions

The assumptions of the study are the same as the ANES assumptions,

which center around the truthful answering of questions by participants,

particularly regarding voting. There have been some well-documented

examples of social desirability bias in political surveys; for example, the

Bradley Effect is the name given to the phenomenon of white voters

overestimating their actual likelihood of voting for a black candidate

(Rogers, 1992). It is possible that some kind of social desirability bias

infected ANES, for example by leading participants to overstate their voting.

Limitations

The limitations of the study are also the limitations of ANES. One

important limitation is that ANES did not survey cell-phone only households

(ANES, 2011)..

Delimitations

The study is delimited to the ANES data about the U.S. Presidential

elections of 1952 to 2008. In addition, the ANES sample was delimited to

only those respondents who gave substantive answers to the questions;

respondents who opted out of answering the main question, of whether or

not they intended to vote, were not included in the final sample.

Significance of the Study

This study is important because it is the first study to apply the full

spectrum of agency theory (including personal agency, social agency, and

sociocultural agency) to the empirical study of voting behavior. Agency

theorists tend to work within their own field of interest; as a result, there is a

dearth of literature that employs each of the three components of agency

theory as a frame through which to examine political behavior. The

methodological significance of the study lies in the use of a quantitative

model, defined and defended at length in chapter three, which can test and

weight the roles of each kind of agency in the prediction of voting behavior.

The study will be of value to a number of audiences. Political

scientists, public policy theorists, political operatives, citizen activists, and

even marketers and psychologists will benefit from an increased quantitative

understanding of the antecedents to voting.

Implications for Positive Social Change

The study will promote positive social change by identifying potential

weak spots in electoral participation, providing a basis for public policy

changes that could increase incentives to vote. The connection to public

issues is as follows: Currently, public resources are applied to get-out-the-

vote efforts on both a federal and state-level basis, but, in the absence of

supporting data, it is difficult to know whether public resources are being

efficiently spent. Identifying the roles of three kinds of agency in impacting

the decision to vote makes it possible to create data-driven get-out-the-vote

strategies, which in turn have a higher likelihood of persuading people to

vote and thus promoting positive political change—on the assumption that

exercising the franchise is a measure of the quality of a democracy. A higher

rate of political participation is directly connected to social change, because

people who are more active in voting tend to feel connected to society in

ways that make them better parents, neighbors, friends, and workers

(Putnam,

2001a, 2001b).

Summary and Transition

This chapter has discussed and provisionally justified a new approach

to answering the question of why people vote, grounded in the ANES data

from U.S. Presidential elections. This approach is based on the use of three

kinds of agency theory—personal agency, social agency, and sociocultural

agency—as part of a single theoretical frame through which to examine

voting behavior. The statistical methods of linear and logistic regression can

be used to determine the relative importance of each type of agency theory

as a prediction of voting behavior, thus generating insights that can help

policymakers to target get-out-the-vote resources in a more rational and

efficient way. Moreover, the use of the three kinds of agency theory is also a

means of generating a richer explanatory model and uniting insights from

theorists who are often not divided by methodological and conceptual

boundaries. Thus, this study can play an important role in mobilizing agency

theory for a more thorough explanation of voting behavior, assisted by the

extensive data in ANES (2011) and the use of regression analysis.

Chapter two offers an overview of the theories used to inform the

study design and that constitute the context of voting behavior. Chapter three

will demonstrate how certain theoretical predictions about why people vote

can be empirically tested by means of a regression analysis of the ANES

data. Chapter four will consist of a presentation and discussion of the

resulting findings. Chapter five will summarize the main findings, discuss

the implications of the findings for past and future researcher, and conclude

with a reflection on the importance of electoral participation in democratic

societies.

Chapter 2: Literature Review

Introduction

The scope of the literature review was broad, as theories of political

participation date from the work of Aristotle to contemporary times. There

was a cutoff date of 2010 used for supporting research in academic articles,

however this cutoff was not used for texts which support the investigation of

agency theory in the scope of voting as knowledge from seminal work on the

topic was determined to be useful in arguing the delineation of the elements

of agency theory which were studied. Review of literature did not uncover

thesis or dissertation work that would be useful in giving a background on

the topics studied. Publications related to political science and sociology

were therefore utilized in order to utilize supporting literature to undergird

the performance of the current study. Studies performed before 2010 were

largely excluded from the review of literature, however this exclusion was

not applied to books on the topic which give support and seminal

explanations of the constructs examined. The literature review strategy

involved entering the following search terms into the Political Science

Complete,

Political Science, Expanded Academic ASAP, Academic Search Complete,

and ProQuest

Central databases:

• “individual agency” and “voting”

• “social network” and “voting”

• “class struggle” and “voting”

• “motivations for voting”

• “voting motivations”

These search strings uncovered several hundred articles, many of which

were read and annotated preparatory to writing the literature review. The

second component of the literature search strategy used the snowball method

suggested by Creswell (2009), in which both the texts and the bibliographies

of chosen articles were consulted in order to identify seminal works and

expand the circle of reading. This process identified several seminal works

(e.g., Aristotle’s Politics) cited in many different empirical studies. When

such works were identified, they were procured, read, and annotated to add

context to the literature review.

This literature review incorporates information from 23 peer-reviewed

journal articles in addition to book-length monographs. The search focused

primarily on books because many political scientists have expanded their

empirical work on voting to book length. In addition, the topics of

individual, social, and historical agency have received more theoretical

attention and empirical analysis in book publications than in journal articles.

One of the rationales for this study was to close a gap in the empirical

literature on multiple factors of agency as predictors of voting.

There is broad consensus in the literature that voting is the result of

choice. While voting is described as an act of choice, the controversy

emerges when trying to decide exactly who is doing the choosing (Kang,

2010). Personal agency theorists argue that the ultimate decision-maker in

voting is the individual; while this individual might be somewhat influenced

by their social group, this perspective argues, a vote ultimately transcends

social influence and represents an individual’s desire to exercise agency in

the world of politics (Hersh, 2012). Sociocultural agency theorists instead

argue that the role of the individual is overshadowed by the role of the social

network that they are embedded in, and that it is ultimately the social

network rather than the individual that chooses (Bergh & Bjorklund, 2011).

Finally, historical agency theorists suggest that both the individual and the

social network are acted upon by the forces of history, especially economics

and class struggle, and that it is the historical moment (rather than individual

or social agency) that determines how and why people vote (Marx, 1972).

Many empirical studies have been written on individual and social agency;

historical agency has not similarly been subjected to empirical analysis, with

its theorists primarily writing highly conceptual arguments.

Previous Approaches to the Problem

The question of why people are motivated to vote has often been

answered through qualitative methodologies. These methodologies have

offered rich insight into the participatory behavior of citizens in democratic

countries, but are structurally unable to answer and resolve certain questions

and disputes. Although several qualitative theories of voting behavior have

been developed, however, scholars have been unable to weigh these theories

against each other to see how they operated together in such a way that they

could both support the performance of voting. Over the past two decades

alone, there have been several qualitative studies emphasizing multiple

aspects of voting behavior:

• Partisanship (including ideology and formal and informal

identification with parties) (Payne, Krosnick, Pasek, Lelkes, Akhtar, &

Tompson, 2010; Poundstone,

2008);

• mobilization (meaning the influencing of the electorate by outside

messages)

(Postman, 2006);

• the overall fabric of civic participation representing the quality and

kind of political participation at the family, friend, and neighborhood

level (Gerber &

Rogers, 2009; Putnam 2001a, 2001b);

• religious identity (Frank, 2004);

• socioeconomic status (Marx, 1972);

• personal psychology, including attitudes to information, authority, and

race

(Simon et al., 2008); and

• demographic facts such as age and gender (Scheff, 1994).

This confluence of findings from qualitative study has led to the

creation of what Geertz (2008) termed a thick description (p. 5) of voting

and pre-voting behavior, in which each of the major theories has some a

priori merit and stands in need of weighing against the others. It is highly

unlikely that any person decides to vote because of a single pathway of

motivation (Geertz, 2008). Now that evidence from other behavioral

research programs has accumulated, the more appropriate conclusion is that

human beings make any purposive and important decision on the basis of a

gestalt of other considerations, conscious and unconscious (Zhang &

Margolis, 2006). There is also evidence of noise in human decision-making,

indicating that some of the variability in decisions cannot be explained at all,

at least by variables that are known to researchers (Westen, 2008). Statistical

analysis such as linear and logistic regression can quantify how decisions are

made, however; the R2 measurement and individual Beta values for

independent variables (IVs) can both be used to quantify the relationship

between any IV (such as mobilization) and the DV of voting behavior. In

theory, quantitative analysis can offer more precise descriptions of voting

behavior than qualitative research.

Each of the variables tracked in this study have been used in prior

research on this topic, but not in the agency-comparative manner utilized in

this study. These variables are: (a) potential voters’ perceptions of how much

people can affect government; (b) potential voters’ perceptions of how

important it is to do what authorities tell them; (c) potential voters’ optimism

or pessimism about the nation’s future; (d) potential voters’ interest in

information about government and politics; (e) potential voters’ families’

interest in politics; (f) potential voters’ household income; (g) potential

voters’ home ownership status; and (h) potential voters’ education—can be

justified because they.

Poundstone (2008) drew upon the variables of potential voters’ families’

interest in politics and potential voters’ education. Lin’s (2010) analysis

touched on potential voters’ perceptions of how much people can affect

government, potential voters’ perceptions of how important it is to do what

authorities tell them, potential voters’ optimism or pessimism about the

nation’s future, and potential voters’ interest in information about

government and politics. Finally, Campbell (2006) examined potential

voters’ household income and potential voters’ home ownership status.

These findings are all evidence that voting is likely an activity that is

influenced by a number of different social and sociocultural factors.

Numerous empirical studies have examined individual, social, and

historical agency. This literature review did not, however, identify any

studies that have attempted to detect the operation of each kind of agency in

a quantitative manner. Indeed, many scholars who approached the problem

from a wide variety of perspectives have noted the insufficiency of existing

methodologies as predictors of observed voting (Abrams,

Iversen, & Soskice, 2010; Alvarez, Levin, & Sinclair, 2012; Aragones,

Giboa, & Weiss,

2011; Bergh & Bjorklund, 2011; Grafstein, 2009; Hersh, 2012; Kang, 2010;

Oswald & Poedthavee, 2010; Panagopoulos, 2010; Urbatsch, 2012). It is

therefore legitimate to suggest a new approach to the problem of predicting

voting, especially an approach rooted in quantitative methodology and

attentive to all three forms of agency mentioned in the previous literature.

Overview of Theories of Agency

Personal agency represents a form of methodological individualism

rooted in the neoclassical framework of Adam Smith (1801) whereas

sociocultural agency has historically been represented by Marx (1972).

Smith represented the trend in modern political thought that privileges the

individual as the source of goal-driven, purposive, and free voting behavior.

Marx stood for the proposition that voting was the outcome of sociocultural

forces that directed the decisions and consciousness of the voter. For Smith,

it was the individual who acts and thereby dictates the nature of society. For

Marx, it was society—actually, history, because history dictates the

evolution of society—that acts and thereby dictates the nature of the

individual. In-between these poles are social agency theorists who accept

Smith’s idea that voting behavior is goal-driven, purposeful, and freely-

chosen, but reject Smith’s idea that it is the individual making the choices.

For social agency theorists, the unit of decision is not the individual but,

rather, some collective: For example, households, neighborhoods, language

groups, religious groups, etc. For Marx, the unit of decision is history itself.

Marx did not believe, as social agency theorists do, that political decisions in

democratic countries reflect the communal choices of groups of people, for

two reasons. First, Marx argued that there were only two groups struggling

throughout history, workers and bosses, and that other groups were

irrelevant. Second, Marx believed that history itself was guiding both the

workers and the bosses towards their interests and positions. Thus, Marx

subscribed to a kind of historical determinism that rejected both the

individual and the collective as potential sources of free, goal-driven

decisions.

Following the conceptual divisions of agency advanced by scholars,

the literature review is divided into four larger sections. The first section

contains a discussion of the evolution of voting. The second section touches

on the frame of individualism. The third section employs the frame of social

agency, which is halfway between Smith’s (1801) pure individualism and

Marx’s (1972) pure social determinism. The fourth section employs the

frame of determinism, and examines sociocultural theories of voting.

Because of the possibility of confusion that can arise when trying to explain

the interrelationships between the frames and the theories, some visual

support would be helpful. Accordingly, Figure 1 lists the three theoretical

frames in the study (individual agency, sociocultural agency, and historical

agency) and demonstrates where four bodies of theory (uses and

gratifications, mobilization, civic agency, and sociocultural agency) fit on

the spectrum:

Figure 1. A diagram showing the study’s theoretical framework and the

relative positions of specific theories.

Individual agency is anchored by Smith (1801) and historical agency

by Marx (1972). There is a bias in modern political thought towards the

acceptance of either individual agency or sociocultural agency as frames for

examining voting behavior; accordingly, as clear from Figure 1, most of the

chosen theories also fit on this side of the spectrum. However, for reasons of

theoretical and historical completeness, it is necessary to acknowledge Marx,

even though Marx’s work is not central to the theoretical model of

this study.

The topic of voter decision-making and psychology is vast, but the use

of the frames in Figure 1 resulted in the narrowing of literature to examples

of seminal work from, for example, uses and gratifications theory and

mobilization theory. The second research strategy was to discover

appropriate scholarly resources by employing the Science Direct, EBSCO

Host, and JSTOR databases. The following search strings were employed in

these databases: Voter theory; voting theory; voting in America; psychology

of voting; and motivations for voting.

Before examining theories through the three frames in Figure 1, it

would be helpful to offer an introduction to voting behavior. The purpose of

this general overview is to trace the historical appreciation of voting

behavior and to illustrate the emergence of individual-centered and socially

centered explanations of voting through the history of political thought.

The Evolution of Voting

The Oxford English Dictionary (2011) defined a vote as follows: “An

indication, by some approved method, of one’s opinion or choice on a matter

under discussion; an intimation that one approves or disapproves, accepts or

rejects, a proposal, motion, candidate for office, or the like” (para. 7). The

word appears to have entered the English language in the late 15th century

from the Latin word votum (vow, wish) and its past participle, vovere (to

vow, desire).

In Western political philosophy, beginning with the history of ancient

Greece, there is every indication that voting was understood more as a civic

obligation rather than a discretionary right of the citizen (Evans, 2010). In

Athens, for example, voting (which was limited to free men) was not

obligatory, but it was a social taboo not to vote. In fact, since Athenian

voting took place publicly and ballots were not secret (Pels, Briquet, &

Bertrand, 2006), strong social and political pressure could be exerted on

non-participants in their process (Cartledge, 2009). Because ancient

Athenian society was small enough for many Athenians to know each other

personally, social pressures to vote were particularly intense (Cartledge,

2009). Voting had a strong significance in Athens for two reasons. The first

reason was a material one. Since Athens only went to war on the basis of

direct voting, and many elections centered on the elevation of officers to

generals, there was a strong incentive for the electorate to make its wishes

felt (Pritchard, 2010). Failure to vote in Athens likely meant that the non-

voter exercised no impact on one of the most important decisions Athenians

could face, that of warfare (Lape, 2010). The second reason for the

importance of voting in Athens was symbolic and philosophical. Missiou

(2011) is one scholar who made an explicit connection between the

philosophy of practice of democracy in Athens. In Greek mythology, the city

of Athens itself was named after the results of a vote. Both Poseidon and

Athena wished to have the city named after themselves, and all the gods—

except for Zeus, who remained neutral—voted on behalf of the two divine

claimants to Athens’ naming rights. Zeus’s abstention resulted in a win for

Athena (Pritchard, 2010).

This story is important in its own right, because it suggests that the

independent and relatively egalitarian city-states of Greece, which later

exerted such great influence on the entire history of Western political

organization, prized the ideas of fairness, consensus, and consultation, even

where gods were concerned (Osborne, 2010). To be sure, neither the rhetoric

nor the reality of voting-based democracy has been ideal; throughout history,

even democratic states have excluded some segments of their populations

from voting.

In American political history, voting—like the study of voting

behavior—has gone through several phases. In the years leading up to the

Declaration of Independence, a number of seminal figures in American

politics wrote of voting as a system of basic self-preservation and self-

interest. In 1766, writing to William Shirley, Governor of the Province of

Massachusetts Bay, on the occasion of British imposition of direct taxes on

the American colonies, Benjamin Franklin (1825) stated that:

…the frontiers of an empire are properly defended at the joint

experience of the body of the people in such empire;--it would now be

thought hard by act of parliament to oblige the Cinque ports or sea

costs of Britain, to maintain the whole navy, because they are more

immediately defended by it, not allowing them at the same time a vote

in choosing members of the parliament…(p. 35).

Franklin thus believed that, at least for the American colonists of the

late eighteenth century, voting was a means of defending themselves from

onerous taxation that threatened to destroy the viability of several colonies,

especially after the major economic crisis brought about by the Bank of

England’s decision to forbid the colonies from minting their own money.

The cry of ‘taxation without representation,’ inspired by

Arthur Lee’s (1775) eloquent A Speech, Intended to Have Been Delivered in

the House of Commons, in Support of the Petition from the General

Congress at Philadelphia, became the popular rally for the incipient

American Revolution, and it was premised on the necessity of voting.

Voting lost its urgency after the success of the revolution. In the 1760s

and 1770s, voting had taken on vast rhetorical importance precisely because

it was denied to the American colonists by the British Empire. After 1776,

when America entered the global community of nations, the issue of voting

diminished in importance and was less frequently evoked as a form of

political salvation (McCaffrey, 2004). In fact, from the period from 1776 to

1789, much of the public discussion of voting in America focused on state-

level voting rather than individual voting, as the nation moved to

amalgamate its colonies into a single federal apparatus at the Constitutional

Convention. Many states sent just a handful of representatives, or sometimes

only one representative, to the Convention, meaning that direct voting was

not a factor in the proceedings; the representatives did not arrive with a

mandate to vote in one way or another based on the vote of their respective

states’ citizens (McCaffrey, 2004). Rather, political elites took on magnified

importance at this time, with defenders of federalism engaging in back-room

bargaining for representatives’ votes.

The contemporary American voter arose as a result of the reforms of the

Progressive Era, including the innovations represented by the secret ballot

and the 17th Amendment. Carson and Jenkins (2011) argued that the debt of

the secret ballot in the late 19th century rendered voter intimidation less

likely. At about the same, the 17th Amendment to the Constitution of the

United States established the direct vote for

Senatorial elections and thus further empowered the individual voter

(Gailmard & Jenkins, 2009). It is within this general historical framework

that it is possible to speak of American voting as both a meaningful and a

common political decision for individuals.

Individual Agency

Several kinds of individual agency have been identified in the

literature. This overview will delve into neoclassical decision theory by way

of uses and gratification theory as well as mobilization theory. As Adam

Smith (1801) is the great ancestor of agency theory as it is applied in modern

Western economic and political though, his work will loom large in the

discussion.

Uses and gratifications theory first emerged in the context of

communications, with the scholars Lazarsfeld (1940) and Herzog (1940)

using the theory to explain why certain members of radio audiences chose

particular shows over others. However, in order to understand the true roots

of uses and gratifications theory, it is necessary to discuss neoclassical

decision theory. Neoclassical theory originates with Adam Smith (1801),

who is credited with introducing and popularizing the use of methodological

individualism as a science of decisions.

Before Smith (1801), scholars who studied human decisions had two

general approaches to the topic: Great man theory and divine providence

theory, both of which have their roots in ancient Greek political theory and

were adopted by Christian theorists during the Middle Ages. Divine

providence theory, as championed by Aquinas (1916), implied that the

decisions of individuals carried less weight than divine providence. Thus,

Aquinas (1916) and other Scholastics argued that the political order was

decreed by God.

Great man theory, which may have its roots in Plato’s (1908) Republic,

assumed that the decisions of certain important personages, particularly

kings, carried disproportionate weight in deciding the state of affairs in war,

culture, politics, and economics.

In The Wealth of Nations, the Scottish humanist and economist Adam

Smith (1801) introduced a new idea: That the decisions of individual

humans collectively dictated the reality of the market (and, by extension, of

politics and other aspects of the collective human condition). Smith was one

of the very few figures in Western intellectual history who made the ordinary

individual the centerpiece of a theory of decision-making. Of course, in the

years since Smith, there has been empirical confirmation of his

methodological individualism in economics (Yonay, 1998). The history of

economics after Smith is, in some ways, a history of the failure of every

theory other than methodological individualism (Ingham, 2008). Centralized

economic systems have failed with predictable regularity and the market has

proven immune to manipulation by any actor, or set of actors, over the long

term (Spiegel, 1991). In the wake of Smith, it is possible to conclude that the

building blocks of economic reality are individual men and women whose

decisions, often made in isolation from each other and in the absence of a

guiding hand, collectively determine the status quo. Given that there is a rich

body of literature treating political decisions as a subset of economic

decisions (see Downs, 1957), Smith’s economic thoughts are highly relevant

to electoral politics.

However, even if Smith’s specific ideas about economic decision-making are

discarded, his formulation of methodological individualism must be

acknowledged as the starting point for modern discussions of voting

behavior.

Methodological individualism was not, for Smith (1801), a generic

decision theory, but rather a way of approaching what he called homo

economicus. For Smith’s account of the invisible hand to be theoretically

coherent, it required some form of methodological individualism as an

anchor. However, in the years after Smith, neoclassical economists and

scholars in other fields influenced by the neoclassical turn began to turn

Smith’s limited concept of methodological individualism into a generic form

of decision theory based on the interrelated concepts of want maximization,

risk reduction, and bounded rationality (Blais, 2000). These concepts are the

direct precursors of both uses and gratifications theory and of other forms of

individual agency that inform voting behavior, so they deserve closer

attention.

Bounded rationality is a theoretical lens in decision theory and

economics assuming that people are rational (that is, want-maximizing, self-

interested beings) but also limited in their rationality, for example by not

knowing everything that needs to be known to make a fully rational decision

(Simon et al.,2008). In neoclassical theory, the ultimate arbiter of rationality

is the market itself (Altman, 2006); for example, if a company stock traded

is priced at $100 a share on the open market, this price is in theory fully

rational in that it reflects the sum total of individual valuations of the stock.

However, the individuals making the decisions to buy the stock are limited

in their knowledge of what might make the stock go up and down. Thus, in

bounded rationality theory, humans are basically rational but limited by their

inability (Wittman, 2008) to discern, process, and act upon the sum of

knowledge relating to complex decisions (such as buying a stock or voting

for a political candidate).

The bedrock of bounded rationality theory, as of methodological

individualism in general, is a commitment to the idea of a free individual

(Udehn, 2001). It is for this reason that methodological individualism is so

closely tied to the philosophy of democracy. Bounded rationality posits a

world of free agents picking and choosing between information in order to

take voluntary want-maximizing actions. While the agent’s free will is not

the only aspect of the decision space—which can also include constraints on

the agent’s free will, such as law and social custom—it is, to bounded

rationality theorists, the most explanatorily powerful aspect of all decisions,

including decisions relating to political behavior.

Despite the fact the bounded rationality was in the intellectual

marketplace of ideas after Smith’s (1801) original 1776 publication of The

Wealth of Nations, it took well over another century and a half for theorists

to rigorously categorize the elements of rationality. An early and influential

attempt at such a categorization was made by

Abraham Maslow (1945), who is associated with his concept of the

hierarchy of needs. Maslow accepted the central thesis of bounded

rationality theory, that people act in order to maximize their wants based on

the information and resources available to them, but he rejected the idea that

all wants were equal to each other in explanatory strength. Maslow argued

that the wants most capable of fulfilling a human being, and therefore the

wants that were most intensely pursued by the mass of humanity, were

higher-order emotional needs. This argument was a radical one when

Maslow made it, because there was a common assumption that the most

human needs had to do with food, shelter, and other physical necessities. It

took the work of Herzberg (1966) to reconcile Maslow’s insights with

previous theorists’ prioritization of physical needs. Herzberg argued that

physical necessities were not in themselves capable of leading to profound

satisfaction; however, when these necessities were absent, people felt

dissatisfaction. Herzberg’s decoupling of the concepts of satisfaction and

dissatisfaction was known as Two-Factor Theory, and has been empirically

verified in both experiments and pseudo-experiments many times. In this

sense, the work of Maslow (1945) and Herzog (1966) is at the foundation of

uses and gratifications theory, which, starting in the 1940s, posited that

people consume cultural artifacts (such as radio shows or political messages)

because such consumption resulted in higher-order emotional and cognitive

gratifications. In of uses and gratifications theory, the methodological

individualism introduced by Smith (1801) and the high importance placed

on primarily emotional needs by Maslow came together to provide a

foundation for explaining why people took purposive action. In its early

manifestation, uses and gratifications theory was used to analyze

consumption of entertainment. However, in the 1960s, numerous scholars

recognized the potential of uses and gratifications theory to predict and

describe political behavior (Katz,

Blumler, & Gurevitch, 1974). These developments were related to Downs’s

(1957) seminal work, which suggested that the kind of agent-driven

decisions reflected in uses and gratifications theory applied equally well to

economic analyses of political decisionmaking. For example, both the

decision framework discussed by Downs and uses and gratifications theory

could explain why people preferred to consume certain political messages

rather than others. In uses and gratifications theory, the idea was that politics

was, like entertainment, capable of generating emotions in the electorate that

could only be fulfilled by taking some form of political action, such as

casting a vote or becoming a party activist. Indeed, as the body of uses and

gratification theory grew in the 1980s, scholars discerned that the gap

between politics and entertainment was closing, at least in America

(Postman, 2006). Since the 1980s, studies have repeatedly shown that people

who are most active in political life are also driven more by emotion than

cognition

(Munson 1993), which was also the case with the radio listeners studied by

Lazarsfeld (1940) and Herzog (1940) in their pioneering application of uses

and gratifications theory to communications.

Applied to voting, uses and gratifications theory assumes a framework

of voter hegemony; in other words, voters act consciously to cast their

ballots because of largely emotional (but also cognitive) satisfactions they

associate with this kind of political behavior. The theory has little room for

determinism (whether social, psychological, biological, or otherwise), and

relies on Smith’s (1801) model of the empowered individual, forging ahead

with want maximization on a voluntary basis.

Mobilization Theory: A More Complex Kind of Agency

Emerges

Mobilization theory is a catch-all name for a number of

theories, with various degrees of similarity to each other, that promote what

might be called a marketing- or influence-oriented approach to political

behavior. Mobilization involves the marshaling of remote forms of

influence, such as advertising and marketing. Prandstraller (2010) defined

mobilization as “market offers aimed at gaining mass approval. Marketing

strategies try to make the product ‘seductive’ so that large numbers of people

will buy it”

(p. 30). Marketing is a complex form of agency because, while acting upon

people’s existing inclinations to act in certain ways, it adds influence and

behavioral modification that muddy the question of whether it is the

individual or the marketing message, or the combination thereof, that is

doing the ‘acting.’

In American elections, billions of dollars are spent to reach the entire

electorate with sophisticated, unremitting, and intense political messages. In

addition to formal campaign spending on advertising, both formal and

informal canvassing (of the kind that occurs when a celebrity endorses a

candidate) are also part of the apparatus of mobilization (Prandstraller,

2010).

In some instances however, these theories overlap. Someone who has

been convinced to vote Democratic by a series of advertisements may

voluntarily leave his or her Republican neighborhood to be among

politically like-minded people, in which case there would be an overlap

between mobilization, uses and gratifications, and civic participations

theories. One way of focusing the discussion is to take up Rogers’ (1995)

model of information dissemination, which establishes a natural boundary

for the role of mobilization theory. Rogers, who is the theorist most closely

associated with what is known as diffusion of innovations theory, argued that

ideas are adopted in five sequential stages: (1) Knowledge, (2) persuasion,

(3) decision, (4) implementation, and (5) conformation. Mobilization theory

fits neatly into the second category, that of persuasion. It is certainly

intended to induce decisions, implementation, and confirmation, but at those

stages mobilization theory gives way to other theories that are superior at

explaining what people do when they have made up their minds.

Mobilization theory focuses on how outside agents try to make up people’s

minds in the first place. A seminal work on this topic is Rosenstone and

Hansen’s (2002) Mobilization, Participation, and Democracy in America,

which is based on a version of bounded rationality that is indebted to Smith’s

(1801) methodological individualism. Rosenstone and Hansen argued that

the best metaphor to explain voting was that of stimulation by elites (p. 8).

True, it is the individual who votes, but only as the end result of a dance with

various interests (including media and political interests) that prime the

pump for participation. It is for this reason that mobilization theory

represents an approach that is further to the right, in Figure 1, of uses and

gratifications theory. While uses and gratifications theory accords more

power of choice to the individual, mobilization theory as articulated by

Rosenstone and Hansen (2002) wish to portray individual choice as

emerging from a web of influence.

Mobilization theorists tend to focus on the impact of media messages

on consumer psychology (Couldry, Livingstone, & Markham, 2007).

Because mobilization theorists’ interests and theoretical frames overlap so

closely with theorists who emphasize individual psychology as a predictor of

voting, discussing them together is appropriate. In this category of theory,

some scholars argue that the influence is equally divided; in other words,

that individuals (voters) and message generators (television stations,

advertising agencies, etc.) exercise mutual influence over each other

(Hutchings, 2003) to create a single and communally accountable system.

Others, while accepting the general existence of such a reciprocal system,

argue that the media has more power to generate opinion and behavior than

the individual has power to shape the direction of the media (Hill, 2003).

Neuman (2007) took an opposite tack, arguing that political emotion was

more in the control of the recipient of the message, but that how such

emotion could be translated into action was vulnerable to manipulation.

Sharan (2002) argued that one of the core components of mobilization

theory is the concept of conditioning. The use of certain stimuli, which can

be administered by the kinds of elites discussed by Rosenstone and Hansen

(2002), is intended to bring about positive responses in the agent. For

example, an individual watching so-called attack ads on television might see

a political in a split-frame with Osama bin Laden, with the hope of the

advertiser being to get the viewer to associate disgust with Osama bin Laden

with disgust for the politician portrayed next to him. However, mobilization

theory also has a cognitive component (Southakker, 1961). In the cognitive

version of this theory, the agent is rationally appealed to by messages.

Sharan and other scholars have documented a split between these two

approaches, with political mobilization in particularly leaning away from

cognitive appeals and more towards the sensationalistic, stimulus-response

pattern of advertising.

There are now more inclusive and sophisticated ways of thinking

about consumer behavior and mobilization. For example, Foxall’s (1999)

behavioral perspective model was an attempt to capture the insights of both

doctrinaire behaviorism and doctrinaire cognitivism. Foxall (1999) did not

part ways with behaviorism and in fact called his theory the behavioral

perspective model (BPM). However, BPM acknowledged that behavior itself

has a rational and voluntary component in addition to the effects of classic

conditioning. Foxall, Maddock, and Leek (2000) preferred to describe

decisional behavior in terms of a larger context that included emotions,

cognitions, and responses to stimuli.

Some theorists have resisted this line of thinking. Morrison, Rimmington,

and

Williams (1998) argued that “Post-modernist consumers will often have

multiple, even contradictory, projects to which they are marginally and

momentarily connected” (p. 174). Kotler (1999), while not embracing this

radical claim, still wrote that while core values are not easily changed,

secondary values can be. The implication is that mobilization is the key

variable to consider when trying to model the behavior of the individual

voter. What different forms of mobilization theory share in common is a

vision of an individual agency that is complicated by outside influence.

From this theoretical ground, it is a short jump to social or civic agency,

which consists of theories that place the collective, rather than the

individual, at the conceptual heart of the decision-making process.

Civic Agency Theory

Although there are many variations of civic agency theory, there is not

necessarily enough of a theory gap between them to justify separating

sections to discuss separate theories. The main difference between such

theories lies not in their theoretical apparatus, but rather in their subjects of

inquiry. For example, some theories focus on the determinative and agentic

role of physical neighborhoods, whereas other theories focus on more

abstract collectives, such as religious groups. Variations on these theories

will be considered in this section.

In some ways, civic agency theory overlaps with purely sociocultural

explanations of voting behavior. Both of these bodies of theory appeal to

what some theorists call demotic explanations. The deme is, in ancient

Greek, a word root that refers to a class of people. For years, evolutionary

biologists and other life scientists have argued that much human behavior

thought to be individualistic is actually demotic (Dunbar, Barrett, & Lycett,

2005). For example, an extremely altruistic person might consider himself or

herself to be acting out of purely individual principles of charity, but such

actions also have a demotic value in that they increase the group fitness of

the charity-giver’s deme. There is therefore a dual significance to human

action, an individual and a demotic significance (Okasha, 2009).

Civic agency theory, broadly construed, holds that political behavior

is demotic. However, the difference between civic participation theory and

the kinds of sociocultural theory examined earlier involves the extent to

which the deme chooses. Sociocultural theorists broadly influenced by

Marxism, as well as sociobiologists and other determinists, do believe that

political behavior takes place in groups, but do not vest these groups with

much power of choice. Classic civic participation theorists, such as Putnam

(2001a), are still methodological individualists and non-determinists in the

neoclassical mold, but they recognize a transfer of agency from the

individual to the deme. Putnam (2001b), for example, suggested that civic

groupings—which could be as simple as a two-person household or as

complex as millions of co-inhabitants of a city—seem to exhibit a kind of

agency that is distinct from, even though it is built out of, individual agency.

In democratic societies, civic participation is held to be voluntary,

whereas in authoritarian societies—following the model in Plato’s (1908)

Republic—the root of civic participation is non-voluntary assignment to

certain spheres of communal life. In recent years, the debut of the Internet

has shifted the terms of this debate, as political scientists have acknowledged

the Internet’s ability to span previously daunting distances between people.

To some extent, the Internet has served as a flattener of civic differences,

allowing people to voluntarily band together in virtual communities for

purposes of political behavior (Gainous & Wager, 2011). This kind of use of

the Internet was prominently on display in the run-up to the 2008 election,

during which Barack

Obama’s campaign employed the Internet as a so-called ‘netroots’ of

activists and wellwishers (Liu, 2010).

The phenomenon of civic democratic participation clearly exists.

However, what is less clear is how to model the behavior of civic groupings.

Recent work suggests that a civic grouping builds opinion consensus as a

means of giving group members incentives to behave in a particular way.

For example, Golub and Jackson (2010) argued that, in a generic decision-

making model, “agents…naively update beliefs by repeatedly taking

weighted averages of their neighbors’ opinions” (p. 112). The individual

embedded in a civic context is constantly interacting with that group. As

Golub and Jackson (2010) suggested, the individual takes the group’s

opinion temperature, and then becomes more likely to conform to the group

on the basis of social cognition and social desirability. Social cognition is

a powerful theory, and one that has often been employed to understand

voting behavior. At its heart, social cognition theory is simply the claim that

people learn by watching others (Bandura, 1997). Both simple and complex

forms of behavior emerge and are reinforced through example. Thus, people

are more likely to take on the beliefs and behaviors of those they have grown

up around, in particular family and friends. In political terms, there is a well-

documented trend of families voting for a single ticket, and sociologists have

also revealed that religious preference and many other forms of behavior

track closely with group affiliation (Bandura, 1986). Social cognition is the

theory that explains this observed phenomenon.

Social cognition explains some of the agentic aspects of civic

participation, as the theory provides a model of how people learn political

behavior in groups, and then how people keep their political behavior

updated by taking continuous stock of the group, as Golub and Jackson

(2010) argued. However, social cognition does not explain why people

would want to immerse themselves in the group in the first place. Social

desirability theory is a sociological theory that fills this explanatory gap.

Social desirability holds (Giddens, 1987) that people obtain significant

pleasure from, and save themselves from distress by, conforming to their

immediate group (such family, neighborhood, workplace, and place of

worship). Thus, the overlap between civic participation theory and uses and

gratifications theory becomes clearer. Individuals obtain gratifications from

the group, which serves as reciprocal reinforcement for group behavior, and

in this sense there is a constant and mutual process of conforming

adjustment between the individual and the group (Scheff, 1994). The group,

meanwhile, becomes a self-sustaining institution that offers rewards for

entry and disincentives for dissent, further cementing the loyalty of

individual adherents.

The theory of partisanship, in the American context, is at heart the

claim that dominant parties create an electorate that is both polarized and

‘big tent’ in nature

(Bogaards & Boucek, 2010; Caplan, 2007; Fiorina, 1999). Since there are

two major parties of roughly equal strength, there are also two polarized

electorates, each of which has begun to code the other as deviant (Andersen

& Taylor, 2005). In this process, partisanship has to some extent succeeded

in turning political opponents into what Cohen (2002) called folk devils.

Scholars have empirically tracked increasing polarization beginning in the

early 1990s (DiMaggio, Evans, & Bryson, 1996; Evans, 2003).

Moreover, since there are only two viable party choices, each party has to

accommodate fellow travelers who might not be in the same party if not for

the perceived necessity to aggregate their votes and achieve political

efficacy. In this way, as Munger (2010) argued, American electoral

democracy harbors a tension between the expression of ideology in the form

of party membership and the suppression of ideology demanded by joining a

party alongside people of essentially different leanings, such as a genuine

socialist alongside a centrist in the Democratic Party. Outside the American

context, partisanship may retain a sense of group survival (Gust, 2008),

whereas in America the fate of no social group is directly tied to party

supremacy.

Religious identity has been held to be a determinant of voting in

recent decades in particular, given the association of the Republican Party

with conservative Christian values (Norris & Inglehart, 2004). Borgida,

Federico, and Sullivan (2009) argued that religious identity is part of the

individual’s political psychology and cannot be excluded from analysis in

any study of the motivation of voting. Brennan and Lomasky (2008) also

argued that religious emotions, cognitions, and heuristics informed political

behavior, especially when one party’s politicians were more able to co-opt

religious identity into part of an alluring form of self-presentation (King &

Leigh, 2009).

Some theorists believe that patterns in American voting can be

understood as part of a global shift in voting dynamics away from what

class-based to culture-based values. For example, a study conducted by

Achterberg and Houtman (2006) revealed that, in the Netherlands, voting is

not determined more by cultural values than by class interests, so that

cultural conservatism and cultural liberalism are the surest predictors of

voting. Achterberg and Houtman conducted a meta-review of literature

revealing that the same pattern obtains in many other Western democracies,

including the United States, and concluded that the data point to a “cultural

explanation” (p. 75) for voting. Such a cultural explanation would subsume

religious values and social values into a single construct.

There may be some overlap between Achterberg and Houtman’s

(2006) cultural explanation of voting and Darmofal’s (2010) self-efficacy

theory . Damofal (2010) conducted a study revealing that people who feel

their vote is more decisive are more likely to vote; it is also the case, as

Achterberg and Houtman (2006) discovered, that selfefficacious voters are

often the ones with the most cultural investment in the outcome of a

decision. In other words, the people who overestimate the prospects that an

election will determine the cultural tone of their society are also the most

likely to vote. While these new theories of voting have evidence to support

them, they do not in themselves resolve the question of whether the decision

to vote is based on individual and agency versus determination.

The common factor in various forms of civic theory is the insistence

on the concept of free and goal-driven decision-making, but the transference

of the center of decision-making power from the individual to the deme.

Whether the deme is represented by the neighborhood, family, or religion,

it—rather than the individual—becomes what social agency theory considers

the prime mover in decision-making.

Historical Bodies of Theory

Historically-oriented theories of voting behavior have been, since

Marx (1972), radical theories that do not accept voting as the product of

actual and legitimate wantmaximization, but rather as the dependent variable

of some pre-existing fact of economic or social organization. The classical

formulation of this theory lies in Marx’s concept of false consciousness,

which after Marx’s death became one of the pillars of Marxist theories of

political behavior and participation in democracies.

The theory of false consciousness suggests that, when anyone but a

bourgeois or someone else with an innate interest in the continuity of a

capitalist state votes, they are under an illusion that they are acting in their

own interest. According to Marx (1972), the interest of the vast majority of

society lay in abolishing the existing capitalist order, and, for workers,

peasants, and other socially and economically disenfranchised people,

participation in this order was contrary to their true wants and needs. In this

way, Marxism—and subsequent theories that have prioritized the importance

of sociocultural factors as determining voting behavior—clashes directly

with what Lukes (1974) called liberalism (best typified by the

methodological individualism of Smith, 1801) and reformism.

Abrams (2010) argued that Marx’s (1972) concept of false

consciousness was the wrong way in which to approach voting behavior in

capitalist democracies. While Abrams adopted the same general theoretical

frame, the argument was that income position is too limited a lens through

which to examine voting behavior. Abrams suggested that a theory of voter

turnout rooted in the voter’s overall position in society was a better predictor

of voting, but did not ground this claim in quantitative analysis.

Other theorists embedded in the sociocultural tradition argue that voting

itself is a sham. Saiegh (2011) argued that genuine rule is achieved by statue

and influence-peddling, a claim that has also been made by Poundstone

(2008). Some theorists point to low American election turnout (Hill, 2006;

Institute for Democracy and Electoral Assistance, 2012) as evidence that

many citizens have reached the same conclusion.

A group of theorists in the sociocultural school de-emphasizes group

economic interests, which were of such concern to Marx (1972), in favor of

more cultural explanations. Dalton and Anderson (2011) argued that

electoral choices are expressions of historical forces as they act upon the

individual. Some theorists in sympathy with Marxist explanations have

pointed out that false consciousness can exist in the field of cultural as well

as economic and political behavior (Frank, 2004). In the cultural

explanations school, Putnam (2001a, 2001b) and Campbell (2006) have been

influential in arguing that, rather than determined per se, voter behavior is

merely shaped by culture and community, a softer claim that is difficult to

reject and that separates some of Marx’s (1972) explanatory power from its

ideological trappings.

Summary of Theories

Attempting to answer the question of why people vote is fraught with

immense difficulties, because possible answers to this question invoke

decision theory, rationality, free will, determinism, emotions, cognitions, the

environment, and social and economic facts. In the welter of all of these

variables, it is just as unlikely that a final theory of voting will emerge as a

final theory of human behavior will emerge. At best, scholars can illuminate

different components of the question, and empirical models can be created to

explain some of the variance in a statistical model with voting as the

dependent variable. The preceding sections of the literature review surveyed

various scholarly approaches that can point the way to a better, but not final

or complete, theory of voting behavior that is tested by the methods utilized

in chapter three.

Chapter 3: Research Method

Introduction

The purpose of this study was to examine the relative contributions

made to voting in U.S. Presidential elections by personal, social, and

sociocultural agency. The contributions of these factors were specifically

explored using information available through the publicly available

American National Election Study (ANES) dataset. A quantitative,

correlative research design drawing on panel data was chosen to determine

the magnitude of these contributions. A quantitative methodology was

appropriate for the research questions of the study because it sought to

measure the contributions of specific independent variables pertaining to

agency to the dependent variable of voting.

The hallmark of the quantitative methodology is a focus on measuring

the impact of one or more independent variables on one or more dependent

variables (Creswell, 2009). Correlative research is research conducted based

on existing data. Sincet would be impractical to scale voting decisions or

agency in laboratory conditions, the correlative approach is the proper means

of collecting data to analyze the research questions of the study. Thus, only

quantitative and correlative methods can effectively support the analysis of

data from the thousands of subjects who need to be sampled in any

statistically representative analysis of the U.S. electorate. The remainder of

the chapter discusses and defends the specific components of research design

associated with this approach.

Research Design and Approach

Regression analysis was applied to ANES presidential election voting

data in order to examine the research question. Regression is a statistical

procedure that can measure the relative contribution of different independent

variables to a particular dependent variable. The independent variables in

this study were personal, social, and sociocultural agency, and the dependent

variable was respondents’ voting. Logistic regression was an appropriate

statistical approach because it is designed to deal with dependent variables

that are dichotomous (Creswell, 2009). Since the dependent variable in this

study has only two possible values (voting or not voting), logistic regression

was applied as the main statistical technique.

Setting and Sample

This study exclusively used secondary ANES data. This panel dataset

was chosen for this analysis because it is the largest dataset on American

voters’ behaviors and preferences. ANES is an ongoing panel study that

began in 1948 for the purpose of tracking American voters’ opinions and

characteristics. In 2008, ANES collected two sets of data: A September 2 to

November 3 pre-election window, and a November 5 to December 30 post-

election window (ANES, 2008). The data in this study were drawn from the

ANES pre-election window for 2008.

Instrumentation and Materials

The purpose of the study was to examine the relative contribution of

the independent variables of personal, social, and sociocultural agency to the

dependent variable of voting. The survey instrument used in this study, the

ANES questionnaire, includes a variety of questions designed to measure the

variables needed to answer the research question. The dependent variable

was measured by a single question that asked ANES respondents if they

intended to vote in the upcoming election. For purposes of this study, only

those respondents who answered yes or no to this question were included in

the sample; those who indicated that they did not know or who did not

answer the question were excluded.

Three classes of independent variables were used in this study:

Personal, social, and sociocultural agency. Personal agency was measured

using variables: (a) How much people can affect government, (b) how

important respondent thinks that people do what authorities tell them, (c) is

respondent optimistic or pessimistic about nation future, and (d) respondent

interest in information about government and politics. These four variables

were chosen because they express generic, non-partisan individual attitudes

about governance that can be used as proxies for the individual’s sense of

political engagement. Social agency was measured using the degree of

respondent’s family’s interest in politics. This variable was the only

identified ANES variable that asked respondents to evaluate the strength of

their social circle’s commitment to political life, and was therefore the only

appropriate variable for capturing social agency in this data set.

Socoiocultural agency was measured through the following variables: (a)

household income, (b) home ownership, and (c) education. Each of these

variables has been shown to motivate the decision to vote (Caplan, 2007).

Table 1

ANES Study Variables

Variable

Type

Question Topic

Personal

Agency

Independent

Belief that individuals have a day in

government

Degree of interest in the elections

Degree of interest in public affairs

Social Agency

Independent

Mother’s party

Father’s party

Sociocultural

Agency

Independent

Respondent income

Respondent home ownership

Respondent education

Demographics

Intermediate

Race, gender, religion, age, ethnicity, major

party affiliation

Voting Dependent Voting in Presidential election

The personal agency variables, social agency variables, and voting

variable were all scored on a Likert scale measuring subjects’ attitudes. For

purposes of analysis, only substantive responses to these questions were

considered; ‘don’t know’ or blank responses were not considered. This

exclusion is supported by DeBell (2013) in his paper on the decoding of the

ANES. The sociocultural agency variables were measured continuously (for

income) and categorically (for home ownership and education). The

demographic variables contained categories from census data; the only

special consideration is that major party affiliation includes people who

reported being Democratic or Republican, while not being registered as

such. For purposes of the analysis, all those who lean toward a party were

counted as belonging to that party.

Data Collection and Analysis

ANES work is performed by researchers at Stanford University and

the

University of Michigan who were working with a National Science

Foundation (NSF) grant. The inclusion criteria for the 2008 ANES data

collection were being a U.S. citizen born on or before November 4, 1990,

having a landline telephone, and living in a U.S. household (ANES, 2008).

The ramifications of this exclusion is that people who fit the demographic of

not having a landline are not represented in the study. The method of

sampling was stratified list-assisted random-digit-dialing (RDD) of landline

telephones. The ANES researchers began by using automated telephone lists

to isolate 51,386 randomly-chosen residential telephone numbers, although

it was not reported by the performers of the ANES survey whether the

randomness was achieved through computer algorithm or manual procedures

(ANES, 2008).

The number of usable cases from the ANES dataset that were used in

this study was over 28,000. Usable cases were those where there was not a

“don’t know or “blank space and where all items were responded to by

respondents. This sample was chosen because it consists of all the

respondents to the 1952-2008 versions of ANES who reported a voting

outcome. A power analysis was performed and the results are given in

chapter four, for the purposes of quantifying the generalizability of the

sample. ANES’s stratification methods and the size of its sample yielded a

high level of generalizability.

To address the research question of this study, several variables were

chosen from the ANES dataset. The data were analyzed by means of logistic

regression of the dichotomous variable of voting, with logistic regression

providing an odds ratio as the measure of an IV’s predictive power over

voting. The research question of the study was as follows: How do the

factors of personal, social, and sociocultural agency influence

American voters’ stated voting in U.S. Presidential elections?

H10: There is no significant relationship between age and voting in U.S.

Presidential elections.

H1A: There is a significant relationship between age and voting in

U.S.

Presidential elections.

H20: There is no significant relationship between gender and voting in

U.S.

Presidential elections.

H2A: There is a significant relationship between gender and voting in

U.S.

Presidential elections.

H30: There is no significant relationship between race and voting in

U.S.

Presidential elections.

H3A: There is a significant relationship between race and voting in

U.S.

Presidential elections.

H40: There is no significant relationship between education and voting

in U.S.

Presidential elections.

H4A: There is a significant relationship between education and voting

in U.S.

Presidential elections.

H50: There is no significant relationship between family income and

voting in U.S. Presidential elections.

H5A: There is a significant relationship between family income and

voting in U.S.

Presidential elections.

H60: There is no significant relationship between religion and voting

in U.S.

Presidential elections.

H6A: There is a significant relationship between religion and

voting in U.S.

Presidential elections.

H70: There is no significant relationship between home ownership and

voting in U.S. Presidential elections.

H7A: There is a significant relationship between home ownership and

voting in U.S. Presidential elections.

H80: There is no significant relationship between interest in the

elections and voting in U.S. Presidential elections.

H8A: There is a significant relationship between interest in the

elections and voting in U.S. Presidential elections.

H90: There is no significant relationship between interest in public

affairs and voting in U.S. Presidential elections.

H9A: There is a significant relationship between interest in public

affairs and voting in U.S. Presidential elections.

H100: There is no significant relationship between a belief that one has

any say in what government does and voting in U.S. Presidential elections.

H10A: There is a significant relationship between a belief that one has

any say in what government does and voting in U.S. Presidential elections.

For purposes of hypothesis testing, the independent variables were

measured in ANES as follows. The personal agency variables of (a) How

much people can affect government; (b) how important respondent thinks

that people do what authorities tell them; (c) is respondent optimistic or

pessimistic about nation future; and (d) respondent interest in information

about government and politics were all measured through a 3point Likert

scale, as was the variable of respondent’s family’s interest in politics.

Household income was measured through a 6-point Likert scale

corresponding to income levels. Home ownership was a yes/no dichotomous

variable, and education was a categorical variable reflecting respondents’

degree of schooling. Since all of these variables were quantified in the

ANES dataset, they were all incorporated into a logistic regression model

conducted in SPSS ™. The results of this regression can further be

personalized into different demographic categories, which in this study were

(a) people of different ethnicities (with the major categories being African-

American, White, Hispanic, and Asian); (b) men versus women; (c)

Democrats versus Republicans; (d) Christians versus non-Christians, (e)

people of different sociocultural categories, and (e) the 35-orolder

population (as of the survey administration date) versus the under-35

population, which were included in order to determine how common

markers of individual and group identity (Kiewiet & Lewis-Beck, 2011)

might mitigate the relationship between agency and voting. The results of

the logistic regression (to be considered significant at a level of p < .05) will

be disclosed and discussed in the next chapter of this study.

There are many ways to operationalize the concepts of personal,

social, and sociocultural agency. This study operationalized each of these

concepts in the manner summarized in Table 1 above. Of particular interest

was the role of these variables in predicting voter turnout. The study

hypothesized that personal, social, and sociocultural variables—into which

the variables listed in Table 1 can be sorted—will indeed serve as predictors

of voter turnout, although it not clear which form of agency will be more

predictive than any other.

Threats to Validity

After more closely analyzing the tens of thousands of phone numbers

randomly selected in the initial samples, non-working, business, and fax

numbers were screened out, and only numbers with a matching listed

residential address were retained by ANES researchers (ANES, 2008). These

households were then sent a letter, with $2 included, explaining the study

and requesting the recruited participants to consent to an initial interview on

the telephone and follow-up monthly surveys (ANES, 2008). Another

potential threat was raised by the procedure of only sampling homes with

landlines, given that many younger people and technology aficionados no

longer have landlines, preferring cell phone and Internet telephony. ANES

acknowledged this possible threat to validity and took one measure against it

by treating voice-over Internet protocol (VOIP) numbers as landlines.

Ethical Procedures

The study was conducted based on publicly available data for which

the original data collectors had IRB approval from their affiliated

universities, as well as from the government of the United States (ANES,

2008). As such, no ethical procedures or concerns were associated with the

data, other than the general ethical directives of analyzing data honestly and

disseminating results. Both of these directives were followed in this study;

all data were double-checked for accuracy and otherwise handled in

accordance with Walden University IRB # 04-05-13-0101627.

Summary and Transition

A quantitative, correlative approach based in logistic regression was

used to determine the extent to which factors of personal, social, and

sociocultural agency influenced American voters’ stated voting in U.S.

Presidential elections. All data were drawn from the ANES (2008) dataset,

whose high number (> 28,000) of subjects, statistically-representative

sampling, and robust study design made it an appropriate source for

information about both agency variables and voting in U.S. Presidential

elections. The results of the analysis are presented and discussed in chapter

four.

Chapter 4: Results

Introduction

The purpose of this quantitative study was to explore how voting

emerges from a combination of personal, social, and sociocultural agency as

measured in the ANES dataset of the American electorate. The research

question was: How do the factors of personal, social, and sociocultural

agency influence American voters’ stated voting in U.S. Presidential

elections? Twelve null hypotheses were associated with this research

question, with each null hypothesis utilizing the dependent variable of voting

in U.S. national Presidential elections and separately tracking one of the

independent variables: age, gender, race, education, family income, religion,

home ownership, father’s party, mother’s party, interest in the elections,

interest in public affairs, and believing that one had any say in what the

government does.

Data Collection

Data were collected consistent with Walden IRB # 04-05-13-0101627.

The ANES dataset was accessed online from October 11, 2013 to February

1, 2014 in order to obtain the data for analysis. During the process of data

collection and analysis, the researcher discovered that data for some of the

variables existed only in certain years of the ANES survey. The implications

of this were that it was not possible to test all variables in both models In

order to fully test the hypotheses, the ANES Time Series Cumulative Data

File was accessed. The ANES Time Series Cumulative Data File contained

data from

1948 to 2008 in a single SPSS file. These data were used to create two

models, one from

1952 to 1988 and another from 1952 to 2008, in order to ensure that the

appropriate variables were represented. These procedures, while deviating

from the initially outlined research plan, did not threaten the quality of the

analysis, for reasons that will be discussed at greater length below.

I originally selected the 2008 ANES dataset on the assumptions that

(a) all of the variables of interest were present in the 2008 data and (b)

previous years of the ANES dataset did not employ the same variables. I

subsequently discovered that two of the variables in my initial study design

were not present in the 2008 data, but that each of the other variables

required by this design existed in the ANES dataset going back to 1952, I

decided to split the analysis into two models, one from which 2008 was

excluded because of the absence of mother and father party identification

variables. I further decided to take advantage of the cross-variable

comparability in the extended data set by adding several more ANES years

to the study.

Descriptive Statistics

The ANES dataset used for analysis included roughly 28,000

individual records of electorally eligible Americans surveyed from 1952 to

2008. Table 2 summarizes the independent, dependent and covariates under

investigation in this study. None of the variables in the study were

continuous (a variable, such as weight, that can be in theory be assigned any

value between its minimum and maximum value [Nikoletseas, 2010]). Each

variable was either categorical and could be sorted into categories that are

not susceptible to any form of ordering or other form of numeric relationship

with each other, or ordinal and had a clear order (Nikoletseas, 2010)). For

this reason, standard descriptions of central tendency (such as mean and

standard deviation) could not be validly calculated for the demographic

variables. For purposes of the inferential statistics utilized in this study, the

inability to calculate measures of central tendency for demographic variables

has no practical implication, as the statistics were chosen on the basis of the

types of data available.

Table 2

Values and Categories of Variables

Variable

Category

Frequency

Age

1 = 17-24

2 = 25-34

3 = 35-44

4 = 45-54

5 = 55-64

6 = 65-74

7 = 75+

9.69%

21.63%

20.64%

16.90%

14.05%

10.51%

6.59%

Gender

0 = Male,

1 = Female

44.14%

55.86%

Race

0 = White

1 = Non-White

79.98%

20.02%

Education

1 = Grade school or less

2 = High school

3 = Some college

4 = College or postgraduate

15.68%

46.35%

20.45%

17.52%

Family Income

1 = 0th to 16th percentile

2 = 17th to 33rd percentile

3 = 34th to 67th percentile

4 = 68th to 95th percentile

5 = 96th to 100th percentile

17.04%

17.73%

32.36%

27.44%

5.43%

Religion

1 = Protestant

2 = Catholic

3 = Jewish

4 = Other and none

64.60%

23.50%

2.38%

9.52%

The majority (53%) of participants were between 18 and 44. The

respondents were also primarily (55.7%) male. For the purposes of this

study, the ANES participants were segmented into two racial categories,

white (79.3% of participants) and non-white (19.9% of participants), with

the remaining 0.8% of participants not having indicated their race. In terms

of religion, the sample included 63.8% Protestants, 23.2% Catholics, 2.4%

Jews, and 9.4% who indicated none or other. In terms of education, 15.5% of

respondents had an education of grade school or less, 45.9% had a high

school diploma, 20.2% had some college without a degree, and 17.3% had a

college degree or postgraduate degree. Income was measured in percentiles;

29.8% of the sample was included in the 34th to 67th income percentile, with

another 25.3% in the 68th to 95th percentile.

The frequencies are very similar in Model 4 as well although there are

a few areas where large differences emerge. One is in race. There are about

5% more non-White voters in Table 4 than in Table 3. This may reflect the

fact that there were socio-cultural reasons for non-White people to not vote

until the late 1950’s-early 1960 with the Civil

Rights Acts of 1957 and 1960 when there were stronger laws put in place to

protect nonWhite voters (Doar, 1997). Education could be reflected upon by

there being greater availability for loans for education, something gained

from policy designed in the 1960s (Hansen, 1983). The role of these

demographic factors in predicting likelihood of voting is discussed later in

this chapter.

Creation of Models

Since the independent variables of mother and party identification

were not collected after the 1988 election, two separate models have been

created: (1) a model in which every independent variable found in the

hypotheses has been included and the years under consideration are 1952 to

1988 and (2) a model from which the independent variables of mother and

father party identification have been excluded and the years under

consideration are 1952 to 2008. Because there are two models, special note

will be made of whether the null hypotheses in the study can fail to be

rejected relative to each of the models. In addition, statistical assumptions

will be tested separately for each of the models.

Statistical Assumptions

One of the main assumptions of logistic regression is that the chosen

predictors are meaningful, with meaningfulness measured by the cumulative

p value of the independent variables (McPherson, 2001). In this study, both

model 1 and model 2 had p values < .001, satisfying the criterion of

meaningful predictors. Another important assumption of logistic regression

is that there is goodness of fit, as measured by R2 or pseudo-R2, in the model

(McPherson, 2001). In both models 1 and 2, the pseudo R2 was over .12,

which indicates that neither model had a high level of goodness of fit.

However, because the focus of the study was on comparing the predictive

power of chosen agencybased independent variables to each other rather

than building a voting prediction model with high goodness of fit, the

relatively low goodness of fit did not compromise the quality of the study.

Another important logistic regression assumption is that there is no

multicollinearity, that is, that two or more independent variables in the model

are not predicted by some combination of other independent variables

(McPherson, 2001). In this study, the results of multicollinearity testing are

presented in Table 5 below. In general, multicollinearity appears when two

or more independent variables are highly correlated; including multicollinear

variables results in redundancy and a loss of model quality

(McPherson, 2011). Multicollinearity is typically tested through variance

inflation factor

(VIF) and tolerance. VIF measures the extent to which the variance of

regression coefficients increase as a result of multicollinearity. Tolerance is

calculated as the reciprocal of VIF.

SPSS has been designed to test multicollinearity only in linear

regression; however, the SPSS help file indicates that this kind of

multicollinearity testing applies to logistic regression as well.

Multicollinearity was tested through both tolerance and VIF statistics. In

keeping with general recommendations about multicollinearity testing, VIF

figures above 10 were held to indicate multicollinearity (McPherson, 2001),

while tolerance figures under .100 were also held to indicate

multicollinearity (McPherson, 2001). By these commonly defined standards,

none of the variables in either model 1 or model were multicollinear. None

of the observed VIF values in either model rose above 2, while none of the

tolerance values in either model were below .5. Thus, based on the existing

criteria (McPherson, 2001), none of the variables in the study were

multicollinear. The tolerance and VIF figures presented in Table 5 below

indicate that both models passed the general assumptions of multicollinearity

in logistic regression.

Table 5

Multicollinearity Diagnostics, Models 1 and 2

Variable

Model 1

Tolerance

Model 1

VIF

Model 2

Tolerance

Model 2

VIF

Age

.809

1.236

.812

1.231

Gender

.946

1.057

.945

1.058

Race

.941

1.063

.939

1.065

Education

.740

1.352

.755

1.324

Family

income

.700

1.429

.693

1.444

Religion

.954

1.048

.963

1.039

Home

ownership

.823

1.216

.810

1.235

Father’s party

.540

1.852

OMITTED

Mother’s party

.544

1.837

OMITTED

Interest in the

elections

.750

1.333

.738

1.355

Interest in

public affairs

.712

1.405

.700

1.429

Belief

about say in

government

.901

1.110

.912

1.096

Results

The results were obtained by performing logistic regression, which

SPSS refers to as binary logistic regression. Logistic regression was used

because the outcome variable had only two conceptually independent states

(voted versus did not vote) that could easily be represented as odds of

voting. Diagnostics and assumption testing for logistic regression have

already been presented. The summarized results appear in Table 6 below.

Table 6

Summary of Models

Note.

Beta

coefficient values of binary logistic regression, standard error in parentheses.

*** = significant at < .001; ** = significant at .05.

Hypothesis Testing

H10: There is no significant relationship between age and

voting in U.S.

Presidential elections.

Variable

Model 1 (1952-

1988)

Model 2 (1952-

2008)

Age group

.181*** (.018)

.220*** (.013)

Gender

.051 (.055)

.088** (.040)

Race

-.032 (.079)

-.008 (.050)

Education

.327*** (.035)

.487*** (.025)

Family income

.205*** (.028)

.213*** (.020)

Religion

.024 (.034)

.016 (.021)

Home ownership

.389*** (.062)

.471*** (.044)

Father’s party

-.016 (.102)

Omitted

Mother’s party

-.027 (.096)

Omitted

Interest in the elections

.496*** (.041)

.553*** (.030)

Interest in public affairs

.143*** (.056)

.199*** (.022)

Belief about say in

government

.359*** (.163)

.355*** (.040)

The first null hypothesis could not be accepted in either model. In

model 1, each unit change in age category was associated with an 18.1%

greater likelihood of voting (p < .001); in model 2, each unit change in age

category was associated with a 22% greater likelihood of voting (p < .001).

As predicted by Putnam (2001a, 2001b) increasing age is associated with

increasing participation in political and civic institutions. That finding was

borne out in both models of this study.

H20: There is no significant relationship between gender and voting

in U.S.

Presidential elections.

The second null hypothesis could not be rejected in model 1 and could

not be accepted in model 2. In model 2, women were 8.8% more likely to

vote (p < .05). The divergence in findings between models can be attributed

to the fact that model 2 encompasses data from later years, in which women

were more likely to vote. Welch (1977) pointed out that female American

voters voted more frequently than men beginning in the 1970s, reversing the

pre-1970s trend of lower female exercise of the franchise. From the 1970s to

2012, women have been observed to participate in American elections at

approximately the same rate as men (Emmenegger & Manow,

2014). These existing empirical findings were borne out in model 2.

The implications of this point will be revisited in chapter five’s

recommendations for practice section. There is, however, literature

suggesting that women are, and have been, more likely to vote than men.

Because women had to fight for enfranchisement, they were historically

more eager to exercise the right to vote (Rosenstone & Hansen, 2002). Even

after women were no longer influenced by the historical memory of

enfranchisement, women’s deep commitment to civic institutions and

communal stability have been cited as reasons for women’s superior

participation at the ballot booth (Rosenstone & Hansen, 2002).

H30: There is no significant relationship between race and

voting in U.S.

Presidential elections.

The third null hypothesis could not be rejected in either model 1 or

model 2, because the p value associated with the variable of race in both of

these models was above .05. Previous analyses of voting patterns have not

disclosed a significant relationship between race and likelihood of voting

(Simon et al., 2008). Simon et al. (2008) noted that, after the fall of Jim

Crow and even in the face of contemporary efforts to make voting more

difficult, African-Americans and other racial minorities have been adamant

about exercising their franchise.

H40: There is no significant relationship between education and voting

in U.S.

Presidential elections.

The fourth null hypothesis could not be accepted in either model 1 or

model 2. In model 1, each unit change in education category was associated

with a 32.7% greater likelihood of voting (p < .001). In model 2, each unit

change in education category was associated with a 48.7% greater likelihood

of voting (p < .001). As predicted in the literature (Poundstone, 2008;

Scheff, 1994), more educated people were more likely to vote in either

model. Poundstone found that Americans in the top 20% of income were

nearly three times as likely to vote as Americans in the bottom 20% of

income. This general trend of increasing voting frequency with increased

education was affirmed in both model 1 and model 2.

H50: There is no significant relationship between family income and

voting in U.S. Presidential elections.

The fifth null hypothesis could not be accepted in either model 1 or

model 2. In model 1, each unit change in family income percentile category

was associated with a 20.5% greater likelihood of voting (p < .001). In

model 2, each unit change in family income percentile category was

associated with a 21.3% greater likelihood of voting (p < .001). As predicted

in the literature (Scheff, 1994), wealthier people were more likely to vote in

either model.

H60: There is no significant relationship between religion and

voting in U.S.

Presidential elections.

The sixth null hypothesis could not be rejected in either model 1 or

model 2. Previous analyses of voting patterns have not disclosed a

significant relationship between religion and likelihood of voting (Norris &

Inglehart, 2004).

H70: There is no significant relationship between home ownership and

voting in U.S. Presidential elections.

The seventh null hypothesis could not be accepted in either model 1 or

model 2.

In model 1, owning a home was associated with a 38.9% greater likelihood

of voting (p < .001). In model 2, owning a home was associated with a

47.1% greater likelihood of voting (p < .001). As predicted in the literature

(O’Sullivan, 2009), home-owners were more likely to vote in either model.

H80: There is no significant relationship between father party

identification and voting in U.S. Presidential elections.

The eighth null hypothesis was tested only in model 1, as the variables

were not available after 1988. The null hypothesis could not be rejected in

model 1. This variable, as well as the variable of mother’s party

identification, does not appear to have been included in previous studies of

voting likelihood, so no meaningful links can be drawn between this finding

and the work of previous scholars. However, there has been general support

(Gerber & Rogers, 2009; Putnam 2001a, 2001b) for the conclusion that

socialization, of which mother’s and father’s voting identification are proxy

variables, conditions individual voting predilections and habits.

H90: There is no significant relationship between mother party

identification and voting in U.S. Presidential elections.

The ninth null hypothesis was tested only in model 1. The null

hypothesis could not be rejected in model 1.

H100: There is no significant relationship between interest in the

elections and voting in U.S. Presidential elections.

The tenth null hypothesis could not be accepted in either model 1 or

model 2. In model 1, every unit change in degree of interest in the elections

was associated with a 49.6% greater chance of voting (p < .001). In model 2,

every unit change in degree of interest in the elections was associated with a

55.3% greater chance of voting (p < .001). Hersh (2012) suggested that

personal interest is the strongest predictor of voting, but did not quantify this

claim.

H110: There is no significant relationship between interest in public

affairs and voting in U.S. Presidential elections.

The eleventh null hypothesis could not be accepted in either model 1 or

model 2.

In model 1, every unit change in degree of interest in public affairs was

associated with a 14.3% greater chance of voting (p < .001). In model 2,

every unit change in degree of interest in public affairs was associated with a

19.9% greater chance of voting (p < .001). Interest in public affairs, or civic-

mindedness, has often been described as a determinant of political behavior

such as voting (Putnam 2001a, 2001b), but the relationship between such

interest and voting behavior does not appear to have been quantified in

previous studies.

H120: There is no significant relationship between a belief that one has

any say in what government does and voting in U.S. Presidential elections.

The twelfth null hypothesis could not be accepted in either model 1 or

model 2. In model 1, every unit change in belief about people’s say in

government was associated with a 35.9% greater chance of voting (p <

.001). In model 2, every unit change in degree of belief about people’s say in

government was associated with a 35.5% greater chance of voting (p <

.001). There is a large body of literature suggesting that individual voters’

belief in the inefficacy of their votes discourages voting (Hersh, 2012;

Kiewiet & LewisBeck, 2011; Munger, 2010; Panagopoulos, 2010; Westen,

2008), but belief in voting efficacy does not appear to have been as widely

studied, especially as a precursor of voting behavior.

Summary of Results

The purpose of the study was to study the factors of agency as

predictors of likelihood of voting. Three main kinds of agency were

considered as predictors: Personal agency, social agency, and sociocultural

agency. Personal agency was measured through subject perceptions of the

following variables: (a) Degree of interest in elections; (b) degree of interest

in public affairs; and (c) belief in individual say over government.

Social agency was measured through mother’s and father’s party

identification. Socoiocultural agency was measured through the following

variables: (a) household income, (b) home ownership, and (c) education.

Based on the empirical results, all of the components of sociocultural agency

(as measured through education, income, and home ownership) and all of the

components of personal agency (degree of interest in elections, degree of

interest in public affairs, and belief that individuals have a say in

government) were significant in both models. Neither of the social agency

variables was significant in model 1, and social agency was omitted from

model 2. When all three types of agencies are considered in one model of

voting, only personal and sociocultural agency predict if someone is going to

vote and measures of social agency were not significant.. The implications

of the results of the study will be considered more carefully in chapter five.

Chapter 5: Conclusion

Introduction

The purpose of this quantitative study was to explore how voting

emerges from a combination of personal, social, and sociocultural agency as

measured in the ANES dataset of the American electorate. Regression

analysis demonstrated that the variables associated with sociocultural agency

and personal agency, but not social agency, were statistically powerful

predictors of the likelihood of voting in presidential elections. This chapter

relates the findings to past theories and empirical findings, discusses the

limitations of the study, makes evidence-supported recommendations, and

discusses the study’s implications for positive social change.

In Model 1 the analysis of data was performed to 1988, however in

Model 2 the analysis was performed to 2008. There is stronger significance

in Model 2 than in Model 1. This finding indicates that in the 20 years

between 1988 and 2008 that sociocultural and personal agency has gained in

strength in order to influence a growth in the effect size. What this means is

that there is an increasing amount of variance that sociocultural and personal

agency are able to explain. This supports the idea that the voting population

is becoming increasingly dependent on the sociocultural and personal

agency factors.

Interpretation of the Findings

The theoretical debate over the nature of agency in voting behavior

has continued for a long time. Although there is a consensus that there are

many independent variables that predict voting (Sexton, 2011), existing

quantifications of various types of agency in the same model are fragmented,

with scholars studying one or a few variables rather than variables that span

all three categories of agency. However, not until relatively recently have

the empirical means of testing these various theoretical positions have

become available. In the United States, the ANES dataset represents a vast

repository of voter data extending back to 1948 and encompassing hundreds

of individual details about the voter.

The idea of voting as an expression of individual agency has been

championed in the neoclassical approach to agency articulated by Smith

(1801). Neoclassical theory suggests that the collective institution of

democracy arises out of the voluntary and purposeful participation of

millions of individual voters, just as the collective institution of the market

arises out of the otherwise disconnected and small-scale buying and selling

activities of innumerable individuals. There are other approaches to the role

of agency in voting. Marx (1972) is closely associated with the idea that

voting is not an expression of agency, but rather the function of structural

forces—in particular, class struggle—that compel individuals to take certain

kinds of actions. The Marxist theory of voting describes American and other

forms of Western democracy as deluding voters into believing that they are

making a free choice. Other political theorists (Gerber & Rogers, 2009;

Putnam 2001a, 2001b) have rejected the Marxist view of agency and also

rejected the neoclassical / liberal version of personal agency, holding instead

that voting is closer to an extension of collective (e.g., familial, communal,

etc.) agency than to personal agency. Widespread support for the existence of

personal agency was found in the models, even after controlling for gender,

income, educational, religious, and racial distinctions. Personal agency has a

significant impact on voting behavior. At the foundation of this work, these

personal agency findings were conceptually related to

Smith’s (1801) idea of individual agency as the driver of collective (i.e.,

economic, political, and social) behavior.

This study adds to the understanding of voting practices of individuals

by looking at the voting behaviors of a larger group of voters in each

election and measuring their voting practices in each period. Several

previous studies have found that people who are interested in elections are

more than twice as likely to vote than people who are not interested (Sanney,

2008; Westen, 2008). People who believe in the efficaciousness of their

individual vote are also at least twice as likely to vote as people who do not

believe in the efficaciousness of their vote (Hutchings, 2003; Rosenstone &

Hansen, 2002). The applicability of these previous findings was, however,

limited by a combination of small sample sizes, delimitation to one election,

and use of dichotomous (e.g., interested / not interest) measures rather than

ordinal or scale measures such as those used by ANES. By using the ANES

dataset, this study was able to address many of the limitations of the

previous studies. The models designed in this study measured how the

factors measured have influenced voting over the past 50 years among a

number of demographic areas.

This study produced several significant findings related to

sociocultural agency. It found that, within the ANES population, a greater

proportion of non-Whites voted in the time period between 1956 and 2008

than 1956 and 1988. This suggests that the voters’ rights laws have had an

effective influence on voting habits of individuals. It also found that voter

education increased over time. This is something that could be tied to the

government increasing the level of funding for education and supporting

loans for students in legislation passed in the 1960s. These results suggest

that public policy may have been the main contributing factor to the growth

of these demographic areas, as variables such as income and gender

remained largely flat during the study period, with little change in

proportions. This suggests that sociocultural agency relies on policy driving

the changes needed for the voting bloc to grow.

These sociocultural agency findings align with numerous previous

empirical findings. These earlier findings were generated using a theoretical

foundation suggesting that individuals’ emplacement within social categories

helps to form their opinions and attitudes, forming voting behaviors (Borgida

et al., 2009; Cwalina et al., 2008; Putnam, 2001a, 2001b). Poundstone

(2008) calculated that Americans in the top 20% of income were nearly three

times as likely to vote as Americans in the bottom 20% of income. Simon

(2008) indicates that Whites and Non-Whites have exercised the franchise in

a statistically identical manner from the 1970s onwards. Emmenegger and

Manow (2014) noted that women vote more frequently than men because of

a greater sense of social responsibility and emplacement. Putnam (2001a,

2001b) found that older Americans voted more frequently, typically because

they are more concerned about protecting their spectrum of interests at the

ballot box. Each of these earlier findings was affirmed by this dissertation

study. These study results also confirm prior research on voting behavior

using different statistical procedures and data models that takes into account

a robust amount of demographic information.

The model tested in this study for 1956-2008 and 1956-1988 does not

indicate that social agency will predict turnout, contrary to the propositions

and findings of prior research such as Putnam (2001a; 2001b). This suggests

that other variables that can be implemented into the understanding of voter

turnout. The use of other variables associated with society and social class

could be used to understand voter turnout in future studies, however the

contributions of Putnam (2001a) that suggest agency being present do not

seem to be valid in this study.

Cumulatively, the results supported the interpretation that both

personal agency and sociocultural agency are significant predictors of

voting. This supports various theories and empirical findings of agency

theory and furnishes useful guidance for policymakers insofar as an

understanding of the factors that motivate voting decisions. These results

also suggest that the American electorate genuinely believes that they are

exercising agency and that their agency matters. While personal agency will

influence a belief that they are voting based on their own beliefs,

sociocultural agency also influences voting. In turn, this finding suggests

that the neoclassical and liberal frames for understanding political behavior,

combined with a healthy respect for the importance of sociocultural

standing, remain highly relevant to American political life.

Limitations of the Study

One of the limitations of the study was that only a few questions in the

ANES dataset were used to measure the concepts of personal agency, social

agency, and sociocultural agency. ANES contains hundreds of individual

questions. Other questions could have been included under the rubric of

personal, social, and sociocultural agency, but were screened out because of

the goal of only including the most relevant ANES questions. This also

prevented the use of the disproportionately high number of highly specific

political questions in the overall ANES data set that only pertain to specific

elections. The questions that were chosen for inclusion in the study did in

fact represent valid types of agency. Another limitation of the study is the

use of self-reported data and the lack of verification measures to determine

whether or not participants’ self-reported voting did not reflect their actual

voting behavior.

Recommendations for Future Research

One of the main findings of the current study is that men and women

differ significantly in terms of what motivates them to vote. Women’s desire

to vote is partially associated with social agency (in particular, degree of

family interest in politics) whereas men have no social agency component in

their voting. In addition, women were highly motivated to vote to the extent

that believed in the possibility of change and also held a personal interest in

politics, while men who believed that it was important to do what authorities

told them to do were more likely to vote.

There are basic differences in how the genders exercise agency. In the

current study, the observed difference between the kinds of personal agency

predictors associated with male likelihood of voting and the kinds of

personal agency predictors associated with female likelihood of voting held

even when income, education, race, party affiliation, and religion were

introduced as moderating variables. Thus, the difference between male and

100

female voting cannot be explained away through a third variable; there

seems to be some aspect of gender itself that is associated with difference in

agency. Some existing research suggests that, in the United States, women

are more socially connected (Frank, 2004), have a greater belief in their

ability to impact public life

(Gerber & Rogers, 2009), and a greater desire to influence positive political

change (Putnam, 2001a, 2001b; Scheff, 1994). These findings could help to

provide a theoretical basis for the gender differences observed in this study.

One possible project of interest to future researchers could be the use

of principal components analysis (PCA) to more precisely identify

differences between male and female voters whose data is in ANES. The

purpose of such a project would not be to explain the origin or the

characteristics of male-female differences in the exercise of agency—a topic

that is more suited to sociology, evolutionary biology, neuroscience, and

related subjects than to political science or public policy—but rather to keep

analyzing the ANES dataset in order to determine whether there are other

indicators of male-female difference that cast additional light on gender

difference in voting preferences. PCA could be run on the entire set of ANES

questions (treated as dependent variables) with the independent variable

101

being gender. Such an approach could facilitate the identification of not

merely isolated differences between male and female voters (such as the

female preference for engagement and the male preference for response to

authority) but rather the existence of larger conceptual categories of

difference emerging from varimax rotation (Creswell, 2009).

Recommendations for Practice

An important aspect of public policy at the federal, state, and local

levels is the development and support of get-out-the-vote programs. In the

first chapter of the study, it was suggested that such programs can only be

efficacious if they act on and engage personal agency. In other words, if

voting is an act that is both personal and voluntary, then appealing directly to

voters’ agency—as so many get-out-the-vote messages do—is bound to get

more results than if voting is the end result of a class struggle or some

manifestation of social agency. The results of the statistical analysis in

chapter four demonstrate that the American electorate is characterized by a

strong sense of personal agency. Therefore, the current approach to get-out-

the-vote messaging appears to be the correct one. Such messages tend to

consist of direct appeals to voters, both invoking and relying upon personal

102

feelings such as pride of participation, the desire to make a difference, and

interest in shaping the country.

Based on the findings of the study, this trend in get-out-the-vote

messaging ought to continue. However, it is possible to further target get-

out-the-vote messages by gender. As the statistical analysis in chapter four

disclosed, women appear to be socially engaged (in terms of having their

voting being more susceptible to family interest in politics) and more driven

by considerations of personal interest than by men, who are

disproportionately influenced to vote by their belief that it is important to do

as authorities ask. Accordingly, it is possible to craft gender-specific get-

out-the-vote messages that suggest that voting is a duty (to men) and that

voting is personally fulfilling (to women). Segmenting get-out-the-vote

messages in this way might lead to improved voter recruitment and

participation, given what was learned about some of the differences between

male and female motivations to vote. However more research on male-

female voting differences is necessary before firm recommendations about

get-outthe-vote messages can be made. Even if such segmentation policies

are not adopted, public policy officials considering get-out-the-vote

103

messaging would be well advised to appeal to voters’ and potential voters’

social agency.

Positive Social Change

On the assumption that exercise of enfranchisement is a cornerstone of

democratic societies, many public policy initiatives focus on convincing

more people to vote. The premise behind such initiatives is that voting is a

voluntary behavior that can be encouraged; in other words, that voting is the

result of personal agency. From this perspective, the main positive social

change implication of the study was the support found for the relationship

between personal agency and voting. Because personal agency is a

significant predictor of voting, get-out-the-vote efforts that focus on building

personal interest in, and commitment to, the political process are likely to be

effective in raising the voting rate, which in turn would represent positive

social change, especially in the context of low voter participation in the

United States (Alvarez et al., 2012).

Conclusion

Raising voter participation in democratic societies levels is a means of

improving the quality of democracy (Alvarez et al., 2012; Aragones et al.,

2011; Newman, 2007; Saiegh, 2009). In order to raise voter participation,

104

public policymakers can rely on a number of methods. One common

approach is to appeal to the individual voter’s sense of interest,

empowerment, and engagement (Cwalina et al., 2008). The main conclusion

of this study is that such appeals are likely to work, because factors of

agency are statistically significant and strong predictors of voting likelihood

in U.S. Presidential elections. In particular, building prospective voters’

interest in elections, interest in public affairs, and belief that they can have a

say in government are all strategies that are likely to increase the likelihood

of voting.

105

References

Abrams, S. J. (2010). Where everyone knows your name: A socio-rational

theory of voter turnout in advanced industrial democracies. New York, NY:

Basic Books.

Abrams, S.A., Iversen, T., & Soskice, D. (2010). Informal social networks

and rational voting. British Journal of Political Science, 41(2), 229-

257.

Achterberg, C. & Houtman, D. (2006). Why do so many people vote

unnaturally? A cultural explanation for voting behavior. European

Journal of Political Research, 45(1), 75-92.

Altman, M. (2006). Handbook of contemporary behavioral economics. New

York, NY: M.E. Sharpe.

Alvarez, R. M., Levin, I., & Sinclair, J. A. (2012). Making voting easier:

Convenience voting in the 2008 Presidential election. Political

Research Quarterly, 65(2), 248262.

Andersen, M. & Taylor, H. (2005). Sociology: Understanding a diverse

society. New

York, NY: Cengage.

106

ANES. (2008). 2008 time series study. Retrieved from

http://www.electionstudies.org/studypages/2008prepost/2008prepost.h

Aquinas. T. (1916). The gospel of stewardship. In W. Hamilton (Ed.),

Current economic problems (pp. 24-47). Chicago, IL: University of

Chicago Press.

Aragones, E., Gilboa, I., & Weiss, A. (2011). Making statements and

approval voting.

Theory and Decision, 71(4), 461-472.

Aristotle. (1912). Nichomachean ethics. New York, NY: Macmillan.

Bandura, A. (1986). Social foundations of thought and action. Philadelphia,

PA:

University of Pennsylvania Press.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY:

McGraw-Hill.

Beaumont, E. (2011). Promoting political agency, addressing political

inequality: A multilevel model of internal political efficacy. The

Journal of Politics, 73(01), 216-231.

Benney, M., Gray, E. P., & Pear, R. H. (2013). How people vote: a study of

electoral behaviour in Greenwich. London: Routledge.

107

Bergh, J. & Bjorklund, T. (2011). The revival of group voting: explaining the

voting preferences of immigrants in Norway. Political Studies, 59(2),

308-327.

Blais, A. (2000). To vote or not to vote?: The merits and limits of rational

choice theory.

Pittsburgh, PA: University of Pittsburgh Press.

Bogaards, M., & Boucek, F. (2010). Dominant political parties and

democracy:

Concepts, measures, cases and comparisons. Hoboken, NJ: Taylor &

Francis.

Borgida, E., Federico, C. M., & Sullivan, J. L. (2009). The political

psychology of democratic citizenship. New York, NY: Oxford

University Press.

Brennan, G., & Lomasky, L. E. (2008). Democracy and decision: The pure

theory of electoral preference. Cambridge, England: Cambridge

University Press.

Campbell, D. E. (2006). Why we vote: How schools and communities shape

our civic life.

Princeton, NJ: Princeton University Press.

Caplan, B. D. (2007). The myth of the rational voter: why democracies

choose bad policies. Princeton, NJ: Princeton University Press.

108

Çarkoğlu, A., & Kalaycıoğlu, E. (2007). Turkish democracy today:

Elections, protest and

stability in an Islamic society. New York, NY: I.B, Tauris.

Carson, J. L. & Jenkins, J. A. (2011). Examining the electoral connection

across time.

Annual Review of Political Science, 14, 25-46.

Cartledge, P. (2009). Ancient Greek political thought in practice. Cambridge,

England:

Cambridge University Press.

Cohen, S. (2002). Folk devils and moral panics. New York, NY: Routledge.

Côté, J. E., & Levine, C. G. (2014). Identity, formation, agency, and culture:

A social psychological synthesis. Psychology Press.

Couldry, N., Livingstone, S. M., & Markham, T. (2007). Media consumption

and public engagement: Beyond the presumption of attention.

Basingstoke, England:

Palgrave Macmillan.

Creswell, J. W. (2009). Research methods. Thousand Oaks, CA: Sage.

Cwalina, W., Falkowski, A., & Newman, B. I. (2008). A cross-cultural

theory of voter behavior. New York, NY: Haworth Press.

109

Dalton, R. J., & Anderson, C. (2011). Citizens, context, and choice: How

context shapes citizens' electoral choices. Oxford, England: Oxford

University Press.

Darmofal, D. (2010). Re-examining the calculus of voting. Political

Psychology, 31(2),

149-174.

DeBell, M. (2013). Harder than it looks: Coding political knowledge on the

ANES.

Political Analysis, 21(4), 393-406.

DiMaggio, P., Evans, J., & Bryson, B. (1996). Have Americans’ social

attitudes become more polarized? American Journal of Sociology, 102, 690-

775.

Downs, A. (1957). An economic theory of democracy. New York, NY:

Harper.

Dunbar, R. I., Barrett, L., & Lycett, J. (2005). Evolutionary psychology: A

beginner’s guide. London, England: Oneworld.

Eligur, B. (2010). The mobilization of political Islam in Turkey. Cambridge,

England:

Cambridge University Press.

Emmenegger, P., & Manow, P. (2014). Religion and the gender vote gap:

Women’s changed political preferences from the 1970s to 2010.

Politics & Society, 42(2),

110

166-193.

Evans, J. H. (2003). Have Americans’ social attitudes become more

polarized? An

update. Social Science Quarterly, 84, 71-90.

Evans, N. (2010). Civic rites: Democracy and religion in ancient Athens.

Berkeley, CA: University of California Press.

Fiorina, M. P. (1999). Extreme voices: a dark side of civic engagement. In T.

Skocpol & M. Fiorina (Eds.), Civic engagement in American

democracy (pp. 395-423).

Washington, D.C.: Brookings Institution Press.

Foxall, G. R. (1997). The emotional texture of consumer environments: A

systematic approach to atmospherics. Journal of Economic

Psychology, 18, 505-523.

Foxall, G. R., Maddock, S., & Leek, S. (2000). Situational determinants of

fish consumption. British Food Journal, 102(1), 18-39.

Frank, T. (2004). What’s the matter with Kansas? New York, NY:

Macmillan.

Franklin, B. (1825). The works of Dr. Benjamin Franklin. New York, NY:

Van Norden.

111

Gailmard, S. & Jenkins, J. A. (2009). Agency problems, the 17th

Amendment, and representation in the Senate. American Journal of

Political Science, 53(2), 324-

342.

Gainous, J., & Wagner, K. M. (2011). Rebooting American politics: The

internet revolution. Lanham, MD: Rowman & Littlefield Publishers.

Geertz, C. (2008). Thick description. In A. Budd (Ed.), The modern

historiography reader (pp. 5-17). London, England: Routledge.

Gerber, A. S. & Rogers, T. (2009). Descriptive social norms and motivation

to vote:

Everybody’s voting and so should you. The Journal of Politics, 71,

178-191.

Giddens, A. (1987). Social theory and modern sociology. Palo Alto, CA:

Stanford

University Press.

Golub, B. & Jackson, M.O. (2010). Naïve learning in social networks and

the wisdom of crowds. American Economic Journal: Microeconomics,

2(1), 112-149.

Grafstein, R. (2009). The puzzle of weak pocketbook voting. Journal of

Theoretical

Politics, 21(4), 451-482.

112

Gust, E. J. (2008). The Arab citizens of Israel: Motivations for collective

action. Ft.

Belvoir, VA: Defense Technical Information Center.

Hersh, E. (2012). Primary voters versus caucus goers and the peripheral

motivations of political participation. Political Behavior, 34(4), 689-

718.

Herzberg, F. (1966). Work and the nature of man. New York, NY: World

Publishing.

Herzog, H. (1940). Professor quiz: A gratification study. In P.F. Lazarsfeld

(Ed.), Radio and the printed page (pp. 64-93). New York, NY: Duell,

Sloan & Pearce.

Hill, D. L. (2006). American voter turnout: An institutional perspective. San

Francisco,

CA: Perseus Books.

Hill, T. G. (2003). Interest-ing candidates: The electoral impact of interest

group endorsements. Columbus, OH: Ohio State University.

Hutchings, V. L. (2003). Public opinion and democratic accountability: How

citizens learn about politics. Princeton, NJ: Princeton University

Press.

Ingham, I. (2008). Capitalism. London, England: Polity.

113

Institute for Democracy and Electoral Assistance. (2012). Voter turnout data.

Retrieved from http://www.idea.int/vt/

Kahanek, J. E., & Eshbaugh-Soha, M. (2009). An informational theory of

midterm elections: The impact of Iraq war deaths on the 2006

election. Denton, TX:

University of North Texas.

Kang, M. S. (2010). Voting as veto. Michigan Law Review, 108(7), 1221-

1281.

Katz, E., Blumler, J. G., & Gurevitch, M. (1974). Utilization of mass

communiction by the individual. In J.G. Blumler & E. Katz (Eds.),

The uses of mass communications: Current perspectives on

gratifications research (pp. 19-32).

Thousand Oaks, CA: SAGE.

Kiewiet, D. R. & Lewis-Beck, M. S. (2011). No man is an island: Self-

interest, the public interest, and sociotropic voting. Critical Review,

23(3), 303-319.

King, A., & Leigh, A. (2009). Beautiful politicians. Canberra, Australia:

Centre for

Economic Policy Research, Australian National University.

Kotler, P. (1999). Principles of marketing. London, England: Prentice Hall.

114

Krugman, P. & Wells, R. (2009). Economics. New York, NY: Worth

Publishers.

Lape, S. (2010). Race and citizen identity in the classical Athenian

democracy.

Cambridge, England: Cambridge University Press.

Lazarsfeld, P.F. (1940). Radio and the printed page. New York, NY: Duell,

Sloan, and

Pierce.

Lee, A. (1775). A speech, intended to have been delivered in the House of

Commons.

London, England: J. Almon.

Liu, B. (2010). The election of Barack Obama: How he won. Basingstoke,

England:

Palgrave Macmillan.

Lukes, S. (1974). Power: A radical view. New York, NY: Macmillan.

Marx, K. (1972). The Marx-Engels reader. Princeton, NJ: Princeton

University Press.

Maslow, A. (1945). The farther reaches of human nature. New York, NY:

Basic Books.

McCaffrey, P. (2004). U.S. election system. New York, NY: H.W. Wilson.

McPherson, G. (2001). Applying and interpreting statistics: A

comprehensive guide. New

York, NY: Springer.

115

Miller, B. L. (2005). Why is the South so conservative?: A Marxian analysis

of alienation, religion, and political ideology among poor Southern

white voters.

Blacksburg, VA: University Libraries, Virginia Polytechnic Institute

and State University.

Milner, H. (2010). The internet generation: Engaged citizens or political

dropouts.

Medford, MA: Tufts University Press.

Missiou, A. (2011). Literacy and democracy in fifth-century Athens.

Cambridge,

England: Cambridge University Press.

Morrison, A., Rimmington, M., & Williams, C. (1998). Entrepreneurship in

the hospitality, tourism, and leisure industries. London, England:

Butterworth-

Heinemann.

Munger, M. C. (2010). Ideology and the theory of political choice. Ann

Arbor, MI:

University of Michigan Press.

Munson, W. (1993). All talk: The talkshow in media culture. Philadelphia,

PA: Temple

University Press.

116

Neuman, W. R. (2007). The affect effect: Dynamics of emotion in political

thinking and behavior. Chicago, IL: University of Chicago Press.

Nikoletseas, M. N. (2010). Statistics for college students and researchers.

New York,

NY: Random House.

Norris, P., & Inglehart, R. (2004). Sacred and secular: Religion and politics

worldwide.

Cambridge, England: Cambridge University Press.

O’Sullivan, A. (2009). Urban economics. New York, NY: McGraw-Hill.

Okasha, S. (2009). Evolution and the levels of selection. New York, NY:

Oxford

University Press, USA.

Osborne, R. (2010). Athens and Athenian democracy. Cambridge, England:

Cambridge

University Press.

Oswald, A.J. & Powdthavee, N. (2010). Daughters and left-wing voting. The

Review of

Economics and Statistics, 92(2), 213-227.

Oxford English Dictionary. (2011). Vote. Retrieved from

http://www.oed.com.

Panagopoulos, C. (2010). Affect, social pressure, and prosocial motivation:

Field experimental evidence of the mobilizing effects of pride, shame,

and publicizing voting behavior. Political Behavior, 32(3), 369-386.

117

Payne, B.K., Krosnick, J.A., Pasek, J., Lelkes, Y., Akhtar, O., & Tompson, T.

(2010). Implicit and explicit prejudice in the 2008 American

presidential election. Journal of Experimental Social Psychology,

46(2), 367-374.

Pececi, J.S., Mooney, A., & Thornborrow, J. (2011). Language, society, and

power. New

York, NY: Taylor & Francis.

Pels, P., Briquet, J. L., & Bertrand, R. (2006). The hidden history of the

secret ballot: Cultures of voting. London, England: Hurst.

Postman, N. (2006). Amusing ourselves to death: Public discourse in the age

of show business. New York, NY: Penguin Group.

Poundstone, W. (2008). Gaming the vote: Why elections aren’t fair. New

York, NY: Hill

and Wang.

Prandstraller, G. (2010). The quaternary entrepreneur. London, England:

FrancoAngeli.

Pritchard, D. (2010). War, democracy and culture in classical Athens.

Cambridge,

England: Cambridge University Press.

Putnam, R. (2001a). Bowling alone: the collapse and revival of American

community.

New York, NY: Simon & Schuster.

118

Putnam, R. (2001b). Making democracy work: Civic traditions in modern

Italy.

Princeton, NJ: Princeton University Press.

Rogers, D. W. (1992). Voting and the spirit of American democracy: Essays

on the history of voting and voting rights in America. Urbana, IL: University

of Illinois Press.

Rogers, E.M. & Shoemaker, F. (1971). Communications of innovations: A

cross-cultural approach. New York, NY: Free Press.

Rogers, E.M. (1995). Diffusion of innovations. New York, NY: Simon and

Schuster.

Rosenstone, S.J. & Hansen, J.M. (2002). Mobilization, participation, and

democracy in

America. New York, NY: Longman.

Rubin, A.M. (1983). Television uses and gratifications: The interaction of

viewing patterns and motivations. Journal of Broadcasting, 27, 37-51.

Saiegh, S. M. (2011). Ruling by statute: How uncertainty and vote buying

shape lawmaking. Cambridge, England: Cambridge University Press.

Sanney, S. N. (2008). The influence of right-wing authoritarianism and

social dominance orientation on motivations for voting. Albion, MI:

Albion College.

Scheff, T. (1994). Microsociology. Chicago, IL: University of Chicago Press.

119

Sexton, R.L. (2012). Exploring macroeconomics. New York, NY: Cengage

Learning.

Sharan, A.K. (2002). Encyclopedia of industrial psychology. New Delhi,

India: Anmol

Publications.

Simon, H.A., Egidi, M., & Viale, R. (2008). Economics, bounded rationality,

and the cognitive revolution. New York, NY: Edward Elgar.

Smith, A. (1801). The wealth of nations. Dublin, Ireland: N. Kelly.

Sosnaud, B., Brady, D., & Frenk, S. M. (2013). Class in Name Only:

Subjective Class

Identity, Objective Class Position, and Vote Choice in American

Presidential

Elections. Social Problems, 60(1), 81-99.

Southakker, H.S. (1961). The present state of consumption theory: A survey

article.

Econometrica, 29(4), 704-741.

Spiegel, H.W. (1991). The growth of economic thought. Durham, NC: Duke

University

Press.

Stevenson, B., & Wolfers, J. (2013). Subjective and objective indicators of

racial progress (No. w18916). National Bureau of Economic

Research.

120

The American National Election Studies (ANES; www.electionstudies.org).

ANES 2008 2009 Panel Study [dataset]. Stanford University and the

University of Michigan [producers and distributors].

Thoms, E. M. (2006). Political efficacy, sociocultural status and age as

predictors of voting behaviors in African American women.

Petersburg, VA: Virginia State University.

Udehn, L. (2001). Methodological individualism: Background, history, and

meaning.

London, England: Routledge.

Urbatsch, R. (2012). The paradox of voting intelligently. Public Choice,

150(3-4), 511-

524.

Van der Straeten, K., Laslier, J. F., & Blais, A. (2013). Vote Au Pluriel: How

People Vote When Offered to Vote Under Different Rules. PS:

Political Science &

Politics, 46(02), 324-328.

Welch, S. (1977). Women as political animals? A test of some explanations

for malefemale political participation differences. American Journal

of Political Science, 21(4), 711-730.

121

Westen, D. (2008). The political brain: The role of emotion in deciding the

fate of the nation. New York, NY: PublicAffairs.

Wittman, T. (2008). Agent-based models of energy investment decisions.

New York, NY:

Springer.

Understanding Presidential Voting Motivation by Factors of Agency

Students also viewed