Download The Tea Party Movement: Right-Wing

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts

Nonpartisan blanket primary wikipedia, lookup

Second Party System wikipedia, lookup

Conservative Democrat wikipedia, lookup

Radical right (United States) wikipedia, lookup

Political parties in the United States wikipedia, lookup

Tea Party movement wikipedia, lookup

Ethnocultural politics in the United States wikipedia, lookup

History of left-wing politics in the United States wikipedia, lookup

Know Nothing wikipedia, lookup

Third Party System wikipedia, lookup

American election campaigns in the 19th century wikipedia, lookup

New Left wikipedia, lookup

Allen West (politician) wikipedia, lookup

Tea Party protests wikipedia, lookup

Transcript
THE TEA PARTY MOVEMENT: RIGHT-WING MOBILIZATION IN THE AGE OF
OBAMA
Joseph DiGrazia
Submitted to the faculty of the University Graduate School
in partial fulfillment of the requirements
for the degree
of Doctor of Philosophy
in the Department of Sociology,
Indiana University
July 2014
Accepted by the Graduate Faculty, Indiana University, in partial fulfillment of the requirements
for the degree of Doctor of Philosophy.
_______________________________
Fabio Rojas, Ph.D.
_______________________________
Arthur S. Alderson, Ph.D.
Doctoral Committee
_______________________________
Timothy Bartley, Ph.D.
May 5, 2014
_______________________________
Patricia McManus, Ph.D.
_______________________________
Robert Robinson, Ph.D.
ii
©2014
Joseph DiGrazia
iii
Acknowledgements
There are many people whose contributions to the successful completion of my dissertation and
graduate career I would like to acknowledge. First, I would like to thank my excellent
dissertation committee: Fabio Rojas, Art Alderson, Tim Bartley, Patricia McManus and Rob
Robinson. My committee provided invaluable help and advice over the years and this project
would not have been possible without them. I would particularly like to thank my dissertation
chair and advisor, Fabio Rojas, for making sociology fun and encouraging me to find my own
intellectual style. I would also like to thank friends and colleagues who helped to make graduate
school more enjoyable, including but not limited to Shiri Noy, Kevin Doran, Karissa McKelvey,
Oren Pizmony-Levy and Jaime Kucinskas. Finally, I would like to thank my family, who were
always supportive and encouraging.
iv
Joseph DiGrazia
THE TEA PARTY MOVEMENT: RIGHT-WING MOBILIZATION IN THE AGE OF OBAMA
The Tea Party movement emerged in early 2009 as a conservative reaction against the election of
Barack Obama and the proposed bailout of financial firms and homeowners by the federal government.
The movement went on to oppose Obama’s health care reform proposal and eventually to support
conservative Republican candidates in the 2010 midterm elections. This dissertation uses the Tea Party
movement as a case study for understanding conservative political mobilization and how conservative
social movements are shaped by other social processes such as political agenda setting, policy formation,
race, the welfare state, and the national media.
This research on the Tea Party brings together a diverse variety of data sources, including survey
and polling data, extensive data on Tea Party events collected through scraping online Tea Party event
calendars and social networking sites, and data on aggregate Google search frequencies. The primary
argument of this dissertation is that social movements like the Tea Party are composed of distinct
constituencies which are mobilized in different ways that depend on fluctuations in the national media
and policy environment. The principal empirical chapter examines predictors of Tea Party mobilization
on the state and county levels over time and finds that local factors associated with mobilization are
dependent on the national media discourse and policy debates at the time. Other findings show that Tea
Party supporters and activists tend to be more conservative and Republican than the general public, are
more likely to fear downward economic mobility and are more likely express hostility toward racial and
ethnic minority groups. Additionally, pre-existing cultural attitudes, such as hostility toward immigrants,
are associated with higher levels of Tea Party mobilization on the state level.
__________________________________
__________________________________
__________________________________
__________________________________
__________________________________
v
Table of Contents
Chapter 1: The Tea Party Movement and Right-Wing Politics…………………..………….…....1
Chapter 2: An Exploration of Novel Internet Data Sources……………………………………..33
Chapter 3: Individual-Level Predictors of Tea Party Support and Activism………………….....53
Chapter 4: Tea Party Mobilization and the National Political Discourse……………...………...84
Chapter 5: Searching for the Tea Party’s Origins: Google Search Data as a Predictor of StateLevel Tea Party Mobilization……………………………………………………………..……125
Chapter 6: Conclusions and Implications………………………………………………………151
References………………………………………………………………………………………167
Vita
vi
Chapter 1: The Tea Party Movement and Right-Wing Politics
The Tea Party movement emerged in early 2009 as a conservative reaction against the election of
Barack Obama and the proposed bailout of financial firms and homeowners by the Federal
Government. The movement went on to oppose Obama’s health care reform proposal and
eventually to support conservative Republican candidates in the 2010 midterm elections. The
Tea Party serves as an interesting case for the study of contemporary social movements,
especially right-wing social movements, as it is the first large scale right-wing populist
movement to emerge in the US in several decades and it is extremely well documented with
polling data, event records and data generated online by internet users. As such, the Tea Party
provides an excellent opportunity for the advancement of sociological theories of right-wing
politics and right-wing political mobilization.
Many scholars have argued that traditional, structuralist theories of social movement
mobilization, namely political process and resource mobilization theory, do not fit conservative
or right-wing movements well (e.g. McVeigh 1999; Van Dyke and Soule 2002; McVeigh 2009).
Conventional social movement theories tend to argue that social movements emerge when
aggrieved constituencies are able to access an opening in a typically closed opportunity structure
and martial the resources of higher status outsiders to their cause (McCarthy and Zald 1977;
McAdam 1982).
However, McVeigh (1999; 2009) argues that conservative movements are not responses
to openings in the political opportunity structure, but rather reactions against such openings,
which high-status groups perceive as threats to the status-quo. Therefore, conservative
movements, whose constituencies tend to be higher status and have ready access to resources and
1
the political system, tend to occur when the political opportunity structure would seem to be
opening for their opponents rather than for themselves. In this sense, most right-wing movements
are reactions against gains by lower status groups. Most explanations for right-wing and
conservative mobilization have focused on the importance of status politics; efforts to restore the
prestige of salient racial, ethnic, religious or cultural identities (e.g. Gusfield 1963), restore or
preserve economic advantages (e.g Brustein 1996), maintain political power, or some
combination of these (e.g. McVeigh 1999; McVeigh 2001; McVeigh 2009). In other words,
explanations for conservative movements have focused on the emergence of grievances among
high-status groups arising from perceived threats to the status quo.
In this dissertation, I propose an explanation for conservative mobilization that
incorporates these threat-based explanations, while also accounting for how right-wing social
movement mobilization is produced when potentially threatening social changes interact with the
larger national political environment. This process is depicted visually below in Figure 1.1.
[FIGURE 1.1 ABOUT HERE]
This model essentially argues that right-wing mobilization is a function of potentially threatening
structural and social changes such as demographic or economic shifts, a perception of a threat
from such changes among high status groups and a political and media environment that
politicizes these threats and makes them salient. When I refer to the political media environment
or the national political discourse, I am referring to the political issues that are dominating media
attention or national debate at a particular time. For example, I will argue that different sources
of threat were salient and different social conditions were associated with mobilization during
the time period when health care reform was dominating the national political discourse (roughly
Spring 2009 through Spring 2010) than during the 2010 election time period (roughly Summer
2
2010 through Spring 2011). The notion that movements alter their tactics, targets, strategies and
mobilization structures over time as the social context in which they operate changes is
consistent with the literature on protest cycles (e.g. Tarrow 1993; Staggenborg 1998; Tarrow
1998; Blee and Courrier 2006; Heaney and Rojas 2011). Additionally, others have argued for the
importance of the national political discourse in politicizing local conditions (e.g. Hopkins
2010). However, none of these approaches have been applied to right-wing social movement
mobilization.
This remainder of this chapter will consist of a brief description and history of the Tea
Party movement, followed by a review of the social movement literature, an outline of the
empirical chapters and a brief discussion of the research designs employed in the empirical
chapters.
A DESCRIPTION AND HISTORY OF THE TEA PARTY MOVEMENT
What is the Tea Party?
The Tea Party movement is a conservative and right-wing populist movement that
emerged in the United States in early 2009 as a reaction against the election of Barack Obama
and legislation proposed in response to the economic crisis that began in 2007 (Zernike 2010;
Ashbee 2011). Generally, the movement is characterized by intense opposition to President
Obama and Congressional Democrats and advocates for drastic reductions in government
spending as well as an originalist interpretation of the U.S. Constitution (Armey 2009; Zernike
2010; Ashbee 2011). At the same time its adherents hold socially conservative views and the
movement is considered by many to be hostile to racial, ethnic and religious minority groups
(Potok 2010). The movement is known for its rallies against the federal stimulus package, health
3
care reform and its support for conservative Republicans, particularly non-incumbents, in the
2010 midterm elections. The movement has also benefited from substantial support from elite
organizations and donors (Burghart and Zeskind 2010; Williamson, Skocpol and Coggin 2011).
The Tea Party Movement can be distinguished from other conservative groups and from
the Republican Party, not so much by its policy objectives or ideology, but by its populist
rhetorical style and identity. As Kazin (1995) argues, populism is defined not by a set of policy
goals, but by language and rhetoric. Populist rhetoric is that which portrays the populists as
patriotic, productive citizens, stuck perilously between amoral, unproductive elites above and a
mass of parasitic, idle poor below. This populist style is evident in the Tea Party’s opposition to
social provision for the poor as well as their rhetorical opposition to Wall Street bailouts and
“establishment” Republicans (Ashbee 2011; Courser 2012). Courser (2012) contends that the
Tea Party is motivated by a perception that neither political party nor the political system is
responsive to their needs, while Ashbee (2011) argues that the Tea party is caught in a tension
between their criticism of establishment members of both parties and their cultural “affection”
for the Republican Party.
The Beginnings of the Tea Party
Although conservative and libertarian activists have been using Tea Party related imagery
and rhetoric for some time (e.g. a reenactment of the Boston Tea Party by Ron Paul supporters in
December of 2007), the Tea Party movement cannot be said to have emerged until February of
2009, shortly after inauguration of President Barack Obama. The first notable anti-stimulus and
anti-Obama protest took place on February 16, 2009. A woman in Seattle, named Keli Carender,
organized what she called an “Anti-Porkulus Protest” against Obama’s proposed stimulus
4
package. The event caught the attention of conservative blogger, Michele Malkin, and was
promoted by the conservative advocacy group Americans for Prosperity. Shortly afterward
similar protests occurred in Denver, CO and Mesa, AZ. However, despite this early activity, the
origin of the Tea Party Movement is commonly attributed to a speech by CNBC commentator,
Rick Santelli, from the floor of the Chicago Mercantile Exchange on February 19, 2009
(Zernicke 2010). Santelli railed against a mortgage assistance plan proposed by President Obama
and promoted a “Chicago Tea Party” protest as a response.
The government is promoting bad behavior…This is America. How many of you people
want to pay for your neighbor’s mortgage, that has an extra bathroom, and can’t pay
their[sic] bills? Raise their[sic] hand! President Obama, are you listening?...we’re
thinking of having a Chicago Tea Party in July. All you capitalists that want to show up
to Lake Michigan, I’m going to start organizing (Williamson, Skocpol and Coggin 2011
p. 26)
Santelli’s rant quickly went viral online and was seen by millions of viewers on online
video sharing websites such as YouTube. Santelli provided energy and excitement for
conservatives who felt disaffected by recent developments, and provided a name and narrative to
the emerging protests. Throughout the Spring of 2009 Tea Party protests continued to occur and
two of the larger Tea Party organizations, Tea Party Nation and Tea Party Patriots, were founded
on April 21 and June 11, respectively (Williamson, et al. 2011). Beginning in August, 2009,
Americans for Prosperity and Freedom Works began to help coordinate Tea Party “town hall”
events in which Tea Party activists attended Congressional town hall meetings to protest the
enactment of the Patient Protection and Affordable Care Act. During this summer, an
organization calling itself the Tea Party Express was established and organized a bus tour across
1
Although the Tea Party Patriots formally registered as a non-profit organization on June 1st, a website for this
organization had already been in existence for several months.
5
the United States encouraging protest and activism against the health care reform bill in
Congress.
Tea Party activism continued throughout the rest of 2010 into 2011. Important events
included the Taxpayer’s March on Washington, organized by the Tea Party Patriots in September
of 2009, the National Tea Party convention, organized by Tea Party Nation in February of 2010
and the New Tea Party Express Bus Tour in the Spring of 2010. Additionally, April 15th saw
many Tea Party “Tax Day” events and protests across the country. The Tea Party remained
highly active throughout the 2010 election cycle and backed a number of conservative
congressional candidates in Republican primaries and in the general election. The movement has
been credited by some observers as being a major force behind the considerable gains in
Congress and statehouses achieved by the Republicans in the 2010 election (e.g. Davis 2010).
Tea Party activity has declined considerably since the 2010 election. A timeline showing Tea
Party activity expressed both by average daily Tea Party patriot events as well as normalized
Google search frequency for the term “Tea Party” between January 2009 and March 2011 is
provided below.
[FIGURE 1.2 ABOUT HERE]
Tea Party Rhetoric and Ideology
The official rhetoric of the Tea Party has focused almost exclusively on economic issues,
with particular emphasis on the deficit and the size of government. In fact, some Tea Party
leaders have gone out of their way to avoid engaging with cultural issues (Zernike 2010). The
official position staked out by Tea Party leaders is a staunch right-wing libertarian position,
concisely summed up by Dick Armey and Matt Kibbe (2009) in their Tea Party Manifesto as “a
6
reaction to what they view as a government that has grown too large, spends too much money,
and is interfering with their freedoms (p. 66).” Additionally, the Tea Party Patriots, the largest
and most influential of the Tea Party groups, declares in their mission statement that “[t]he
impetus for the Tea Party movement is excessive government spending and taxation (“Tea Party
Patriots,” n.d.).”
Despite the Tea Party’s claims that its purpose is to advocate libertarian fiscal policy on
behalf of those who feel their interests are threatened by the expansion of government, many
commentators have argued that much (or even most) actual Tea Party support is driven by
anxiety rooted in xenophobia, racism and fear of cultural difference and change. Those who
argue that the Tea Party is largely fueled by these sentiments often contend that the election of
the nation’s first African American president, rather than anger over policy, served as the catalyst
for the emergence of the movement (Potok 2010). This perception has been fueled by the
presence of racist and xenophobic signs and reports of the use of racist slurs at Tea Party rallies.
Additionally, cursory examination of polling data seems to indicate that, while Tea Party
supporters are certainly more fiscally conservative than the average American, they largely do
not conform to the staunch libertarian views of some Tea Party leaders like Armey. For example,
according to an April, 2010 NYT/CBS poll, the majority of Tea Party supporters favor large
government programs like Medicare and Social Security.
Groups Involved in Tea Party Activism
Tea Party Nation
Tea Party Nation was organized in April of 2009 by a Nashville, Tennessee attorney
named Judson Phillips along with his wife Sherry Phillips (Burghart and Zeskind 2010). The
7
group, organized as a for-profit company, runs a social networking site and organizes local and
national Tea Party events (Vogel 2009). Tea Party Nation is best known for organizing a widely
criticized and controversial National Tea Party Convention in February of 2010. The event took
place in Nashville, Tennessee and attracted a crowd of about 600 paying attendees. The price of
the event, at approximately $550 per attendee, plus an additional $349 to hear Sarah Palin give a
keynote speech, combined with the Tea Party Nation’s status as a for-profit company fueled
resentment and skepticism about Phillips’ motives among other Tea Party members (Associated
Press 2010; Good 2010a; O’Brien 2010; Vogel 2010). The fact that Sarah Palin was rumored to
have received $100,000 as a speaking fee only served to aggravate such tensions further (Vogel
2009).
In the lead up to the event, criticism of the manner in which Tea Party Nation and the
Phillips’ were handling the event’s finances became so intense that several prominent speakers,
including Michele Bachmann (R-MN) and Marsha Blackburn (R-TN) dropped out citing
concerns about the event’s cost, finances and House ethics rules (Associated Press 2010; Jonsson
2010). A group called the American Liberty Alliance, one of the key sponsors of the event, also
withdrew in the midst of the controversy. The controversy, also, added to the perception among
some, that the event was an example of “astroturfing” by corporate interests and the Republican
Party (Good 2010b).
Tea Party Patriots
The Tea Party Patriots is the largest and most grass roots of all the major Tea Party
organizations having a large membership and a comparatively small budget (Burghart and
Zeskind 2010). Tea Party Patriots was founded in the Spring of 2009 by Jenny Beth Martin,
8
Mark Meckler and Amy Kremer and consists of a broad network of local chapters across the US
as well as a social networking site linking local chapters and allowing them to easily
communicate and coordinate events. As of September, 2011, the Tea Party Patriots claim to have
nearly 3000 local chapters, each of which organizes local meetings and events2.
In September of 2009, the Tea Party Patriots was influential in organizing the September
2009 “March on Washington” event, in which several tens of thousands of individuals engaged
in protest against the federal government, the Obama administration and the healthcare reform
proposal in Congress. The Tea Party Patriots also attempted to insert themselves directly into the
policy making process, by backing a manifesto called the Contract From America. The Tea Party
Patriots, with the support of Freedom Works (see description below) used a website to take
suggestions on policy priorities from members who then voted on their top priorities. The final
list of the top ten priorities, all of which focused on reducing government spending and taxation,
were released on April 15th, 2010 (Davis 2010; Jonsson 2010). A large number of Republican
candidates for office and elected officials signed onto the document, including governors and
gubernatorial candidates, senators and representatives.
Tea Party Express
The Tea Party Express, founded in 2009, is a political action committee that raises money
and organizes bus tours in support of Republican candidates for office (Burghart and Zeskind
2010). Unlike other Tea Party organizations, the Tea Party express does not have a membership
nor does it organize and support local Tea Party groups. In this sense, the Tea Party Express is
more akin to an elite fundraising organization than the other Tea Party groups.
2
This figure is based on data compiled by scraping the Tea Party Patriots website on September 16, 2011.
9
The Tea Party Express has had a tense relationship with other, more grass roots, Tea
Party organizations. In late 2009, The Tea Party Express recruited former Tea Party Patriots cofounder Amy Kremer, who had just been ousted from her organization, to become the new chair
of the Tea Party Express. This move resulted in friction between the two groups culminating
with the Tea Party Patriots filing a lawsuit against Kremer demanding that she return control of
the Tea Party Patriots’ website and intellectual property (Tucker 2011). The tension between the
two groups continued to escalate with further legal action when Kremer later sued the Tea Party
Patriots for defamation relating to comments allegedly posted on the social networking site,
Facebook, by fellow Tea Party Patriots co-founder Jenny Beth Martin and her husband Lee
Martin.
In addition to its legal battles and tensions with other Tea Party groups, the Tea Party
Express was also involved in a major controversy over racist comments made by its initial vice
chairman, Mark Williams (Burghart and Zeskind 2010). Williams, who wrote an open letter to
NAACP President Banjamin Jealous containing what was widely viewed as a racist caricature of
African Americans, resigned from the organization shortly after its publication (Kennedy 2010).
Freedom Works
Freedom works is a conservative non-profit group started in 2004 that organizes and funds a
variety of conservative campaigns. The group was founded by former U.S. Representative Dick
Armey, as well as Jack Kemp and Boyden Gray out of a merger between David Koch’s group,
Citizens for a Sound Economy, and a group called Empower America (Armey, Kemp and Gray
2004).
10
THEORY
Social Movement Theory
Much of the early work in sociology on protest and social movements focused on, or was
informed by, right-wing movements and World War II era fascism. Given the highly destructive
and frightening nature of these movements, research focusing on them often looked to sources of
irrationality and psychological or emotional pathology as the basis of such movements. Indeed,
work from this perspective is, perhaps, epitomized by Adorno, Frenkel-Brunswik, Levinson and
Sanford’s (1950) work on the “authoritarian personality.” These researchers argued that certain
sets of psychological characteristics predispose individuals to joining fascist movements.
Additionally, other theories of collective action and activism emerged in the post-war years that
focused on the role of stress, strain, frustration, aggression and deprivation (Davies 1962;
Smelser 1963; Feierabend and Feierabend 1966; Gurr 1970). Essentially all of these perspectives
viewed protest and other forms of political instability as the result of frustration and strain
brought about by social changes. Participants in protest were viewed as being isolated and
anomic, cut off from the larger political and social institutions around them. Others such as
Lipset (1960) and Lipset and Raab (1978) expounded upon this tradition, arguing that right-wing
and conservative movements were primarily a form of backlash against threats to the social
status of formerly high-status groups. These movements were thought to adopt prejudiced and
aggressively nationalistic positions as a way of elevating their status above outside groups.
Theories emphasizing pathological personality types, frustration and irrationality seemed
plausible and appealing when applied to fascist and extremist movements that precipitated
widespread suffering and disastrous wars. However, as the horrors of the Word War II era began
to fade and scholars in the U.S. increasingly began to focus their efforts on the civil rights
11
movement and other progressive social movements of the 1960s, they began to move away from
highly individualistic, psychological theories that focused on irrationality and emotion. Instead,
social scientists began to view social movements as a rational form of political action. Social
movement mobilization was increasingly seen as the result of openings in the structure of
political opportunities within a society and the availability of resources to drive and sustain a
movement (McCarthy and Zald 1977; Jenkins and Perrow 1977; McAdam 1982).
Resource mobilization theory essentially argues that the accumulation of resources by a
movement is central to its ability to successfully mobilize (McCarthy and Zald 1977). The theory
argues that movements are able to mobilize successfully when they are able to extract money and
time from the population, especially from well-off and elite outsiders. According to political
process theory or political opportunity theory, as it is otherwise known, social movements tend to
emerge when there are shifts in the structure of political opportunities that allow the movement
to gain leverage over the political process (McAdam 1982; Tarrow 1998). Such opportunities can
consist of friendly political regimes coming into power, emerging weaknesses or divisions
among elites, or the availability of new resources. At the same time other scholars began to
explain social movement recruitment and participation as a function of social networks (Snow,
Zurcher and Ekland-Olson 1980; McAdam 1986; Gould 1991). These scholars argued that
individuals tend to be drawn into protest activity through their network ties and that movements
tend to recruit most successfully through networks.
Social movement research, for several decades, continued to focus largely on progressive
social movements and primarily focused on the role of social structure in explaining the
emergence and outcome of social movements. As structural explanations for social movements
gained increasing empirical support, the “mass society” approaches that portrayed social
12
movement activity as the irrational actions of isolated individuals came to be seen as obsolete
and largely discredited. However, in recent years, conservative and right-wing social and
political movements have been becoming increasingly prominent in Western Europe and the
United States. In Europe right-wing nationalist parties that promote jingoistic and xenophobic
nationalism along with neo-liberal economic programs have been growing in popularity and
influence (Betz 1993a; Betz 1993b). Additionally, in the United States the Christian Right and
more recently, the Tea Party have been among the most prominent social movements of recent
decades. As McVeigh (2009), in his study of the Ku Klux Klan in the 1920s, has pointed out,
many of the social movement theories built around the civil rights movement and other
progressive social movements do not seem to apply well to conservative and right-wing
movements.
McVeigh contends that conservative movements or movements that aim to preserve or
restore the privileged positions of relatively advantaged social groups, have different motivations
and do not face the same obstacles as movements built around disadvantaged groups. Progressive
movements are seen as being fought by, or at least on behalf of, powerless or socially
marginalized groups who challenge elites and the status quo. The constituents of these
movements are in a constant state of grievance, but usually have little access to resources or
institutional political power. On the other hand, socially privileged groups generally have ready
access to resources, do not face repression and have some access to conventional political
channels. Given that this is the case, social movements comprised of relatively privileged groups
would not seem to be as reliant on the emergence of new political opportunities or the
availability of outside resources, which limits the usefulness of political opportunity or resource
mobilization theory as an explanation for conservative mobilization. For example, Tarrow (1998)
13
argues that large amounts of labor unrest during the Great Depression occurred in the United
States and France, as compared to Germany and Britain, because labor friendly regimes came
into power during that time period in these countries. The Tea Party, on the other hand, emerged
at a time period where the political opportunity structure would seem to be closing for
conservative groups rather than expanding. The 2008 elections saw the Democratic Party take
control of the presidency and consolidate its control of both houses of Congress.
Additionally, Hardisty (1999) argues that the right has long had access to substantial
support and resources made available to it by wealthy individuals and organizations and has built
up a formidable movement infrastructure since the 1970s, but those alone are not enough for
mobilization. In other words, the right has consistent access to resources and political influence,
however exogenous shocks in the form of economic, political or cultural threats to the status quo
are required for mobilization.
While political opportunity and resource mobilization theory are often insufficient as
explanations for why conservative movements emerge, a return to the mass society approach also
seems inappropriate, especially in light of recent research indicating that members of even very
radical and socially marginalized right-wing groups are not socially isolated nor irrational
(Brustein 1996; Blee 2002). For these reasons, the study of conservative and right-wing
movements calls for separate theoretical treatment from left-wing and progressive movements.
Conservative social movements are generally seen as being caused by different social
forces and having different dynamics than leftwing movements. Conservative movements tend to
be reactionary movements – movements that involve high status groups seeking to protect their
privilege against real or perceived social changes and threats. Previous research has argued that
the origin of these threats can be economic, cultural or political, or some combination of the
14
three. Cultural threats are those that are seen as threatening the prestige and cultural dominance
of high status groups. This can include changes in demographics through immigration or internal
migration, changes in religious sentiments or changes in social values. Economic catalysts to
right-wing movements occur when relatively privileged groups see a threat to their economic
interests or security emerge as a result of demographic changes or changes to the national or
global economy. Political threats are those that threaten the political influence of high-status
group, whether through demographic changes that alter the electoral map or through changes in
policy.
Thus, existing theory essentially views the potential members and supporters of
progressive movements as being members of low-status or marginalized groups and, as such,
having relatively constant grievances over time. According to resource mobilization and political
process theory, these grievances are only translated successfully into mobilization when these
groups manage to access usually scarce resources and seize rare opportunities in a typically
closed political opportunity structure. The potential members and supporters of right-wing
movements, on the other hand, have relatively constant access to resources and political
influence on account of their higher social and economic status. However, these are only
translated into mobilization when there exists the perception of a serious threat to the status quo.
This relationship is summarized below in Table 1.1.
[TABLE 1.1 ABOUT HERE]
Explanations for Conservative and Right-Wing Populist Movements
Cultural Threat and Status Politics
15
Lipset and Raab (1978) provide a theory of conservative movements that focuses on
status politics, but is still very much steeped in the notions of irrationality and alienation that
characterized the mass society perspective.3 Lipset and Raab essentially argue that during
periods of intense social change or breakdown, groups that perceive a threat to their status will
attempt to defend their status by seeking associations with groups viewed as having been
prestigious in the past. This identification with past high status groups often manifests itself in
bigotry and xenophobia. Similarly, others have argued that conservative movements are an
irrational manifestation of a desire by certain groups to halt their declining social status or to
elevate their low status (Hofstadter 1967).
Other researchers have argued from a Weberian status politics perspective, taking an
approach that views conservative movements and their participants as being more rational and
goal oriented than Lipset and Raab allow. Gusfield (1963) argues that the US, compared to
Europe, has achieved relative consensus on economic issues and that the primary arena of
struggle has been cultural. Specifically, groups struggle over the allocation of prestige. He
contends that most conservative movements in the US have been attempts by groups to restore
and defend the prestige associated with their own lifestyles and culture that are losing prestige to
or are threatened by new comers. He uses class struggle as an analogy, arguing that in the same
way economic conflict is a struggle over the allocation of resources, status conflict is a struggle
over the allocation of prestige. Given that prestige is socially meaningful and can have tangible
consequences for groups, Gusfield sees the struggle over its allocation as largely rational.
More recently, scholars have emphasized the importance of cultural politics and status
groups in explaining conservative opposition to welfare state expansion and social provision
3
It is worth pointing out that the emphasis on irrationality even forms the basis of the title of their book The
Politics of Unreason.
16
(Quadagno 1990; Skocpol 1992; Orloff 1993; Evans 1997; Gilens 1999; Skrentny 2002;
Steensland 2006). In this way, cultural threat is closely tied to economic issues. Essentially this
literature argues that much of the opposition to progressive programs or programs that benefit the
poor is due to these programs being seen as benefiting disliked or undeserving social groups.
These “cultural categories of worth” are largely based on racial, ethnic, religious and gender
identities.
According to Skocpol (1992), the first US welfare programs began as pension programs
for civil war veterans. These pensions were seen not as a social right but as a reward for service
during the civil war. Later, social scientists and unions attempted broaden these programs into
more expansive social protection and insurance schemes for working class men. One of the
reasons these attempts were unsuccessful is because the initial welfare schemes were founded on
a distinction between the deserving and the undeserving (veterans and non-veterans). In this
sense, working men were not seen as a group deserving special protection or a reward. In
Skocpol’s view, the establishment of the US welfare state as being based on “special claims” has
led to a situation in which the provision of benefits is tied to the notion of deservingness rather
than social rights, leading to public opposition toward benefits to social groups deemed
undeserving. For example, Gilens (1999) has argued that racial attitudes are powerful predictors
of support for welfare spending. This is due to the fact that welfare programs are perceived by
the public as primarily benefiting African Americans and opposition to welfare is therefore
driven to a large extent by anti-black attitudes.
Furthermore, Steensland (2006) specifies several mechanisms through which cultural
categories of worth influence policy development. Specifically, the deeply engrained distinction
between the “deserving” poor and the “undeserving” poor serves to influence policy
17
development through their contribution to collective schemas, their use by actors in deliberation
and in public discourse and by the reinforcement of the boundaries between these categories
through their institutionalization in existing policy. According to Steensland, the barriers
between the deserving and undeserving poor are largely demarcated along racial lines.
Research on cultural threat as a source of right-wing mobilization is not limited to the
United States. Studies of far-right nationalist parties in Western Europe have demonstrated the
importance of xenophobia and cultural categories in driving the emergence and increasing
strength of these parties (Betz 1993a; Lubbers, Gijsberts and Scheepers 2002; Rydgren 2005;
Rydgren 2008; Oesch 2008). These scholars have found that xenophobia and anti-immigrant
sentiment are important predictors of the membership in far-right parties and of the electoral
success of such parties. Oesch (2008) has gone further demonstrating that the anti-immigrant
sentiment that forms the basis of support for these parties is based on xenophobia and discomfort
with pluralism rather than economic concerns about competition with immigrants over wages
and jobs.
Although the Tea Party's rhetoric is focused primarily on economics and their leaders
specifically caution organizers and activists against discussing cultural issues (Zernike 2010), the
notion that status issues and cultural politics are a major driving force behind the movement is
not implausible. The presence of protest signs expressing racially motivated opposition to
government policies and to President Obama at Tea Party rallies were a topic of heated
discussion in the media and in the popular discourse in 2010. Additionally, the Tea Party has
openly feuded with the NAACP and the Southern Poverty law Center contends that, although
“[t]he Tea Parties and similar groups that have sprung up in recent months cannot fairly be
18
considered extremist groups,…they are shot through with rich veins of radical ideas, conspiracy
theories and racism (Potok 2010).”
Additionally, examination of public opinion polls shows that despite the small
government rhetoric espoused by many of the movement’s supporters and activists, most Tea
Party supporters are in favor of many large government programs. For example, according to an
April 2010 New York Times/CBS poll, 62% of Tea Party supporters reported thinking that the
Medicare and Social Security programs are worth the cost, compared to the only modestly
higher, 76% for the general population. Just 33% of Tea Party supporters thought that these
programs are not worth the cost. According to the same poll, 52% of Tea Party members
(compared to just 28% of all respondents) felt that “too much has been made of the problems
facing black people.” Additionally, 73% of Tea Party members reported opposition to the
government provision of “benefits to poor people” compared to only 38% for the general public.
It is particularly interesting to note that while Tea Party members are supportive of universal
entitlement programs that are not associated with any social groups perceived to be
“undeserving” they express strong opposition to programs aimed at helping the “poor” and
minorities. As cursory as this examination is, it is suggestive of a cultural element in the Tea
Party’s small government rhetoric.
Economic and Political Explanations
Although, there seems to be a longer tradition of cultural explanations of conservative
and right-wing activism, recent years have seen a number of economic explanations emerge, as
well. Many observers have pointed out that high unemployment rates are associated with the
emergence of right-wing movements (e.g. Jackman and Volpert 1996). Other scholars have
19
crafted more sophisticated theories of how the rational pursuit of economic interests translates
into the emergence of right-wing groups. Given the Tea Party’s focus on economic issues,
particularly the deficit, in their rhetoric and official messages, investigating economic
explanations for Tea Party mobilization is important.
Brustein (1996) emphasizes the role of economic rationality in right-wing movements,
arguing against so-called “irrationalist” explanations of the rise of the Nazi party. He uses
membership data from the Nazi party before 1933 to show that the membership during this time
span largely consisted of groups who would be expected to benefit most from the Nazi’s
economic policies. He constructs an “interest-based” model to explain membership, arguing that
most of those who joined the party did so out of rational self-interest. Specifically, people joined
when their interests were congruent with the economic policies of the party and the selective
incentives that could be obtained by joining were greater than the costs (i.e. others in one’s social
network were expected to join as well). Brustein goes on to argue that the Nazi party played
down its racist and xenophobic views during the interwar period when it built up its power.
Unlike modern far right groups which attempt to build membership based on racist and
ultranationalist appeals, the Nazis built their membership through rational economic appeals and
then implemented their racist and nationalist plans after obtaining power. For the Tea Party
movement, rational-choice based economic approaches would lead to the hypothesis that more
affluent individuals would support the movement at higher rates than others, seeing increases in
social spending or taxation as a threat to their economic position.
McVeigh (1999, 2001, 2004, 2004b, 2009) focuses on the role of economic and political
threat, while downplaying cultural or status politics as primary motivators of right-wing
activism. McVeigh proposes what he calls a “power devaluation” model of conservative social
20
movements. Essentially, this model argues that conservative mobilization is most effective
among populations that are experiencing simultaneous declines in their “purchasing power” in
both economic and political exchange. The causes of this decline can include changes in the
mode of production, changes in electoral law and demographic changes. When this situation
occurs, groups facing decline will take action to restrict the supply of others who can offer the
same political and economic goods, while stimulating demand for what they have to offer. Often,
cultural identity is invoked as a political weapon to accomplish this end. For example, a group
facing such a decline might use slogans or rhetoric encouraging the purchase of products from
their identity group rather than others (e.g. “buy American.”). McVeigh uses the case of the Ku
Klux Klan in the 1920s as evidence of his theory.
In reality the distinction between economic, political and cultural explanations is often
blurred and the different forms of threat may well frequently be intertwined. Seemingly cultural
movements may have real economic consequences, and cultural attitudes and resentments are
often injected into economic debates. This is evident in racial opposition to the welfare state
where racial and ethnic resentments get tied to economic frustration. Additionally, there are
cases where it may be difficult to determine whether threats associated with outgroups are
cultural or economic in origin. For example, there has been debate over whether opposition to
immigration is rooted in xenophobia and discomfort with cultural changes or whether it is rooted
in the perceived economic competition associated with immigration (Oesch 2008; Rydgren
2008).
Social Movements in a Changing Political Environment
21
Although previous research indicates that theories focusing on political opportunity and
resource mobilization do not fit conservative movements well, changes in the political
environment are still likely to be important in understanding conservative movements. A number
of scholars have argued that movements alter their tactics, targets, mobilization structures,
strategies and alliances over time as the social context in which they operate changes (Tarrow
1993; Staggenborg 1998; Tarrow 1998; Blee and Currier 2006; Heaney and Rojas 2011). These
scholars have argued that changes in the political opportunity structure are important not just
because they provide opportunities for mobilization (as previous literature focusing on
progressive movements as found), but also because they often cause movements to shift their
focus, strategies and even constituencies. For example, Blee and Currier (2006) find that
movements alter their goals and strategies in response to new political opportunities. When new
policies are proposed or elections take place, movements often see these occurrences as
opportunities to effect change or attract new members. As such, they may alter their strategies or
even their goals in order to capitalize on the new opportunity. Other scholars, have found that
political events like elections can mobilize or demobilize movements by altering their perception
of the threat posed by the status quo. For example, Heaney and Rojas (2011) argue that the antiwar movement was demobilized by the election of Barack Obama in 2008.
After its emergence in early 2009, the Tea Party was active through the health care
reform debates of 2009 and 2010, through the end of the 2010 midterm elections. Over this time
period the political environment changed considerably. The Tea Party was confronted with a
number of potentially threatening policy proposals as well as opportunities to influence the
political process. During the earliest days of the movement, shortly of Rick Santelli’s rant on the
floor of the Chicago Mercantile Exchange, the movement seemed to be focused on various
22
pieces of proposed legislation designed to bailout homeowners and financial institutions.
Although some of the more controversial legislation, such as the Troubled Asset Relief Program
(TARP) was actually signed into law under Obama’s predecessor, George W. Bush, it is no
surprise that Tea Party anger over the issue did not emerge until after the election of Barack
Obama. Heaney and Rojas’ (2011) research would suggest that the perceived threat posed by this
legislation would only become highly salient for Republicans after the election of Obama, a
Democrat.
In late 2009 and 2010 the national political discourse was dominated by a health care
reform plan proposed by the Obama administration, a potentially deeply threatening policy
proposal for Tea Party supporters, which first became the focus of Tea Party mobilization during
the so-called “town hall protests” during the summer of 2009. Opposition to health care reform
continued to dominate the Tea Party’s agenda up through the spring of 2010 when the
Affordable Care act was signed into law. Shortly after the passage of the affordable care act, the
Tea Party shifted its focus to congressional primary elections taking place in the late spring and
early summer of 2010 followed by general elections in the fall. These elections provided the
movement with an opportunity to influence the political process and required another shift in
focus and strategy.
This dissertation will address the issue of how right-wing mobilization responds to a
changing political environment, by studying how the forms of threat associated with Tea Party
activism change as the national political environment and media environment change.
OUTLINE OF THE EMPIRICAL CHAPTERS
23
This remainder of this dissertation will include a methodological chapter and three
empirical chapters that address three major research questions pertaining to the model depicted
in figure 1.1. Chapter 2 consists of a methodological chapter that provides a discussion of some
of the novel data sources and methods used in this dissertation, including web scraping and
aggregate internet search data.
Chapter 3, the first empirical chapter, examines who supports the Tea Party and why.
What are the demographics of the Tea Party and what sources of threat (i.e. economic, political,
cultural/status) are salient among Tea Party supporters at different points during the movement?
This chapter examines the differences between Tea Party activists and supporters as well as
changes in the Tea Party’s base of supporters over time. The findings show that Tea Party
supporters tend to be high income, conservative whites who express high levels of animosity
toward racial and ethnic minorities and hold socially conservative views. Tea Party activists
differ from supporters in that they are more likely to indicate that they feel that their economic
status is threatened. The first empirical chapter also shows that, over time, the Tea Party becomes
smaller, more homogeneous and more partisan. This chapter establishes that the Tea Party is
made up of relatively high-status conservatives who express attitudes consistent with the
perception of threats to their cultural and economic status and thus establishes the existence of
“perceived threat” as shown in Figure 1 on the individual level.
The second empirical chapter addresses the question of how structural factors, through
the perception of threat, affect Tea Party mobilization. In other words, how do social conditions
translate into mobilization? Additionally, how do the factors associated with Tea Party
mobilization change as the national political environment and media discourse change? This
chapter uses an analysis of county and state-level data on Tea Party Patriot events as well as
24
national-level data on political discourse based on Google search and Fox News article counts to
address these questions. The findings indicate that the local conditions associated with threat
change over time and are dependent on changes in the national media discourse. For example,
this chapter demonstrates that the effect of county and state-level immigration is higher when the
national discourse is focused more heavily on healthcare reform. This chapter argues that this is
because a national discourse dominated by a discussion of an expansion to the welfare state (a
highly racialized concept among American conservatives) made the threat of immigration more
salient. This chapter established the effects of “social changes” and “public/media discourse” as
shown in Figure 1.1.
Using data on Aggregate web search activity from Google, the final empirical chapter
demonstrates that pre-existing anti-immigrant sentiment and job-seeking behavior are significant
predictors of Tea Party mobilization at the state-level. The fact that the measure of antiimmigrant sentiment is taken from the time period just before the emergence of the Tea Party
movement indicates that pre-existing cultural attitudes towards immigrants were significant
predictors of mobilization even controlling for actual observed immigration rates. Additionally,
the job seeking measure constructed from the search data is a superior predictor of mobilization
than observed measures of unemployment.
RESEARCH DESIGN
This dissertation will involve the analysis of a number of data sources, measuring Tea
Party activity, public discourse and attitudes and economic and demographic factors at both the
individual and aggregate-level. Analyzing data on both the individual and aggregate levels will
25
allow for an understanding of both the contextual and individual-level factors and processes
associated with Tea Party support and activism.
Data
Survey and Polling Data
Individual-level data used in this analysis will come from a number of sources including
a 2010 NYT/CBS News poll that focuses on the Tea Party Movement and includes an
oversampling of Tea Party activists and supporters. This data set will allow for an analysis of
both Tea Party supporters and activists. Additionally several political surveys collected by the
PEW organization dating from February 2010 through June 2011 will be used. This data will be
employed to examine changes in the Tea Party’s base of support over time.
Aggregate Data
An aggregate measure of Tea Party activity and mobilization to be used in this
dissertation is a record of all Tea Party Patriot events that were recorded on the organization’s
event calendar between January 2009 and August 2011. I have recorded 6,193 events from this
time period by periodically scraping the site using an automated script written in Python. Some
of the early Tea Party activity in February of 2009 is missing from the data as the website was
not up and running at that point in time and those events were not retroactively added to the
calendar. These events have been merged with a county-level data set to allow for both countylevel and state-level analysis4. Also included in both the county and state-level data sets are
measures of demographic and economic characteristics from the County Characteristics, 2000-
4
The dependent variable is plotted over time Figure 2.
26
2007 dataset compiled by the Inter-University Consortium for Social and Political Research
(ICPSR) as well as the Bureau of Labor Statistics and the U.S. Census Bureau. Additional
variables include measures of unemployment, economic activity, immigration and population.
Media and Internet Search Data
Throughout this dissertation I employ several measures of national political discourse over time.
I use two data sources to construct these measures of national political discourse: counts of
articles posted on the website, FoxNews.com and aggregate national Google search frequency.
These variables are intended to measure the extent to which the national political discourse was
focused on a given topic over time. I selected Fox News as a media source due to the Tea Party’s
well documented connection to this news source. I also use the more novel data source, national
Google search frequency, to construct measures of how interested the public is in a particular
issue at a given time. These measures represent the relative frequencies with which users in the
United States search for a particular term on the search engine, Google. A great deal of recent
evidence suggests that search frequencies are valid and reliable measures of public interest and
attention (e.g. Stephens-Davidowitz 2014; Swearingen and Ripberger 2014). More details on the
construction and validity of these measures is provided in the individual chapters in which they
employed.
In addition to using national Google search frequencies over time as a measure of
national political discourse, I also employ state-level cross-sectional Google search frequencies
as measures of state-level attitudes and behavioral patterns. The next chapter presents a method
for obtaining these measures and discusses issues relating their validity.
27
Methods
Several different methods of analysis are used in the course of this dissertation project.
The analysis of the individual level data will use conventional cross sectional regression
techniques, specifically logistic regression and multinomial logistic regression models, to model
participation in and support for Tea Party activity. This analysis will provide descriptive
information on the demographics and make-up of the Tea Party with respect to both supporters
and participants. It also provides information on the ideology of Tea Party members as well as
their attitudes toward issues such as race and the economy. The individual-level analysis also
provides insight into the factors that motivate participation in and support for the movement. For
example, the individual level analysis can identify statistically significant associations between,
sources of threat, such as anti-black or anti-immigrant attitudes and Tea Party participation
controlling for other factors. The analysis of the individual level data also includes the analysis
of data at several different time points over the course of the movement in order to assess how
the dynamics of Tea Party support change over time.
The analysis of the aggregate county-level data involves the use of negative binomial
regression models. Negative binomial regression allows for the modeling of count outcomes,
where overdispersion is present (Long 1997). To model mobilization across multiple time points
(e.g. during the health care reform debate and the 2010 election), negative binomial models with
clustered standard errors and time interaction effects are used. This allows for tests on the change
in effects over time. Independent variables in this section include demographic and economic
variables relevant to cultural and economic threat, in addition to variables relevant to the partisan
affiliation of the county. State-level models will use time serious cross sectional models with
months nested in states. These models will interact state-level variables associated with various
28
forms of threat with national discourse variables to ascertain the extent to which the effect of
these local conditions is mediated by the national media discourse.
29
Table 1.1: Structural differences between progressive and right-wing populist movements.
Progressive Movements
Right-Wing Movements
Availability of Resources
Intermittent
Constant
Access to Political Power
Intermittent
Constant
Grievance/Threat
Constant
Intermittent
30
FIGURE 1.1: Theoretical model
Perceived Threat
Cultural
Social Changes
Resources/Organizational
Capacity
Political
Demographics
Economic
Mobilization
Outcome
Elections/Policy
Economy
Public/Media Discourse
31
Figure 1.2: History of the Tea Party represented as Tea Party Mobilization
32
Chapter 2: An Exploration of Novel Internet Data Sources
The rapid growth of new communications platforms on the internet in recent years has made
many new forms of data available to social researchers. In many instances, this data is similar to
data already available to social scientists and the internet simply makes it easier to access and
compile. For example, many organizations archive their activities in documents or databases
available to the public through the internet. Additionally, large archives of news articles allow
researchers to easily find and code large numbers of news articles much more quickly than
would have been possible in the past. There are other forms of data that could be said to be
indigenous to the internet – the traces of online activities and communications that would not
exist without new communications and networking platforms. These types of data, including
social media data from sites like Facebook or Twitter as well as aggregate web search data from
services like Google, provide researchers with insight into the public discourse as well as popular
opinions and attitudes.
As indicated in the previous chapter, this dissertation will make use of several novel data
sources obtained from the internet. These include data on search frequencies from the popular
search engine, Google, as well as data on Tea Party Patriot events obtained through webscraping
the Tea Party Patriots event calendar. This chapter will briefly discuss the motivation for using
this data, the methods used to obtain them and issues relating to their validity.
The Internet as a Social Science Data Source
In recent years the rapid increase in internet use, particularly social networking services and
search engines has generated a tremendous amount of data on the attitudes and behavior of
33
internet users. Unlike data actively collected by a researcher through interactions with research
subjects (e.g. surveys), this data is generated by internet users in the course of normal online
activities and as such, many scholars believe, provides insight into the attitudes of users that is
more candid than information that could be obtained through more traditional means of data
collection such as surveys or interviews. This is especially true for potentially sensitive issues
like attitudes toward race and gender.
Some scholars have expressed concerns about the validity of data produced by internet
users on the grounds that such users are not necessarily representative of the general population
or that the content produced by internet usage is frivolous and not well suited toward studying
the types of phenomena social scientists are typically interested in (e.g. Java, Song Finin and
Tseng 2007; Naaman, Boase and Lai 2010; Barberá and Rivero 2013). However, other research
has found that internet and social media data can be used to produce valid metrics of the offline
world if the process by which the data is produced is taken into account and the metrics are
carefully designed with this process in mind.
For example, research has shown data produced by users of the microblogging site,
Twitter, to be a poor measure of public attitudes due to the non-representative nature of Twitter
users (Mitchell and Hitlin 2013). However a large number of studies have found that data from
Twitter does seem to provide a valid measure of public attention and public interest in topics and
that such measures are correlated with and can even reliably forecast real world events. For
example, it might not be possible to produce precise estimates regarding the percentage of the
population that has a favorable view of a politician, but it may be possible to estimate whether
the public is more interested in one politician than another.
34
Previous research has successfully employed Twitter data to establish measures of public
interest. Asur and Huberman (2010) have found that Twitter sentiment can predict the financial
success of commercial films, Bollen and Zeng (2011) have found that data from Twitter can
predict fluctuations in the stock market, and others have found that Twitter activity correlates
with mundane behaviors such as sleep, work and mood (Dodds, Harris, Kloumann, Bliss and
Danforth 2010; Golder and Macy 2011). Twitter has also been shown to be a reliable metric of
various political behaviors including elections and voting (e.g. Tumasjan, Sprenger, Sandner and
Welpe 2010; DiGrazia, McKelvey, Bollen and Rojas 2013; Huberty 2013) as well as
participation in social movements (Agarwal, Bennett, Johnson and Walker 2014).
While Twitter has been shown to be a good indicator of public interest in issues, another
source of internet data that has proven useful to scholars is aggregate data on search patterns
from search engines like Google. Some scholars have shown Google search data to be an
effective measure of public interest or attention in much the same way that data from Twitter has
been used in this fashion. Perhaps one of the best known uses of search data in this capacity has
been the use of real time search data to predict outbreaks of infectious disease such as influenza
(Brownstein, Freifield and Madoff 2009; Carneiro and Mylonakis 2009). The premise of this
research – and the tools that have been developed for this purpose – is that in areas where a flu
outbreak is emerging or in progress there will be a surge in searches for terms related to the flu
(e.g. regarding symptoms, treatment, health care availability). These techniques have been found
to be more effective in quickly identifying flu outbreaks than traditional epidemiological
techniques1.
1
Note: For a recent criticism of Google Flu Trends see Lazer, Kennedy, King and Vespignani (2014). The authors
argue that recent ‘algorithmic’ changes to Google Flu Trends and statistical problems in estimation have
undermined the accuracy of GFT and caused it to substantially overestimate the prevalence of the flu compared to
CDC records in recent years.
35
Other scholars have found Google search data to be an effective indicator of other social
processes. For example, Vosen and Schmidt (2011) test the ability of Google search data to
forecast consumer spending and find that national-level time series measures from Google search
data outperform the Michigan Consumer Sentiment Index and the Conference Board Consumer
Confidence Index. Additionally, Choi and Varian (2012) show that Google search based metrics
can effectively forecast near-term economic indicators such as sales and consumer confidence.
Other scholars have applied Google search data to political behavior. Swearingen and Ripberger
(2014) show that candidates in US congressional elections who received a larger share of search
traffic than their opponents were more likely to win, net of other factors.
Some scholars have argued that given the properties of Google search, that it can do more
than other data sources to gauge public attention and public interest. Unlike many forms of social
networking and social media in which individuals publicly broadcast their activity, Google
searches generally take place in private and are not visible to the public at the individual-level.
As such, Google searches are likely to be more candid and suffer less from desirability bias than
other forms of social media like Twitter. Stephens-Davidowitz (2014) has found that Google
Searches for racial slurs correlated negatively with Obama’s vote share in the 2008 presidential
election relative to that of Kerry in 2004, arguing that Google searches for anti-black slurs
effectively measure local anti-black sentiment which might make voters less likely to support a
black candidate. Google search data, therefore, may be particularly effective at measuring certain
attitudes that are viewed as socially undesirable and difficult to capture in surveys and thus may
be able to provide good measures of the types of attitudes and behaviors generally associated
with racial and economic threat.
36
Google Trends and Associated Technical Challenges
In the 5th chapter of this dissertation, I construct several state-level measures based on Google
search data from a service called Google Trends based on search terms related to Tea Party
mobilization. The terms employed, “immigrants,” “illegals,” “jobs,” and “guns” are meant to
measure interest in immigration, anti-immigrant sentiment, economic distress and conservative
values, respectively. Google Trends works by providing users with data on the relative, percapita, search frequencies for search terms in a specified region over a specified time frame. It
provides time series data on the frequency of the search term for the whole region as well as
aggregate (cross-sectional) data on the subregions. For example, if the region of interest is the
United States, time series data will be provided for the entire United States and aggregate crosssectional estimates will be provided for each individual state.2 The estimates produced by each
query are based on a randomly selected batch of searches, so each query might produce slightly
different results due to random sampling error.
The data provided are normalized such that, for the time series data, the highest
frequency for any place and time is scaled to 100 and all other values are scaled relative to 100.
The relative frequency of different terms (or combinations of terms) can be compared by
simultaneously including both in the same query (up to five terms can be included in a single
query). When multiple search terms are included, the highest value for any of the terms at any
point in time is scaled to 100 and all other values are scaled relative to that. The cross-sectional
subregional estimates are scaled so that the highest subregional value for each term is 100
regardless of the number of terms included. Thus, the overall prevalence of each term relative to
the others can only be ascertained from the time series data for the entire region.
2
Google trends only supports subnational units smaller than states to a limited extent. Data is not available at the
county-level and is available for only an incomplete set of metro areas.
37
A key technical challenge to acquiring subregional or subnational (e.g. U.S. states)
Google search estimates is that data are only made available for any geographical area when an
undisclosed minimum absolute search volume exists in that region.3 This minimum value is high
enough that many words do not meet the minimum threshold for some US states. For example,
when querying the term “illegals” – a term used to construct a measure of attitudes toward
immigration in a later chapter - over a two year time frame, only 11 state-level aggregate
estimates are provided. The other 39 states had absolute search volumes for that term that were
too low either because of low population or low frequency of use. In order to perform a statelevel analysis that includes all states I have developed a solution to this limitation as described in
the following section.
Proposed Approach
Relatively little work has been done that attempts to develop state-level or subregional Google
search frequency estimates. Stephen-Davidowitz (2012) developed an approach to obtaining
Google search frequency estimates at the media market-level. Although, the description of his
algorithm is too vague to replicate, it is a complicated process that involves resampling from
Google Trends thousands of times to produce each set of estimates. The method employed here
is similar in principle to that employed by Stephens-Davidowitz, but much simpler and easier to
implement than his procedure, and seems to produce estimates that correlate very highly with
observed values. The method proposed here requires only a small number of samples to produce
each set of estimates and produces valid estimates for all 50 states, whereas StephensDavidowtiz’s approach did not produce valid estimates for all media markets.
3
https://support.google.com/trends
38
While Google Trends does not provide state-level cross sectional estimates for terms that
do not meet the minimum threshold, it will provide cross-sectional estimates for joint searches
(searches that include one word and/or another word; this is achieved by using the “+” symbol to
separate the words) even if one term in the joint search does not, individually, meet the
threshold. The approach employed here, like that employed by Stephens-Davidowitz, capitalizes
on this feature by using an arbitrary word that meets the minimum absolute threshold in each
state as a baseline. A query is made for both the baseline word and for searches that include the
baseline word OR the word of interest. The search volumes for the baseline word alone are then
subtracted from the combined volumes.
Equations 1 through 5 describe the procedure I use to produce estimated search volumes
on the state level, followed by an explanation of the procedure as well as illustrations of its
properties using an observable test case. To begin, three samples,𝑠, are taken that include search
volumes for a baseline term, 𝑉𝑠,𝑐 and a combined search volume for the baseline term and the
term of interest 𝑉𝑠,𝑐𝑤 .
The observed search volumes for the word of interest, 𝑉𝑠,𝑤 , are equal to:
(1)
𝑉𝑠,𝑤 = 𝑉𝑠,𝑐𝑤 −
max(𝑉𝑐 )
𝑉 + 𝜖𝑠,𝑤
100 𝑠,𝑐
(2) 𝜖𝑠,𝑤 = 𝑉𝑠,𝑐&𝑤 + 𝜌
The estimated search volumes are, therefore, given by equation 3:
(3) 𝑉̂𝑠,𝑤 = 𝑉𝑠,𝑐𝑤 −
39
max(𝑉𝑐 )
𝑉
100 𝑠,𝑐
Where 𝑉̂𝑠,𝑤 , represents the estimated search volume for the word of interest, 𝑤, and max(𝑉𝑠,𝑐 ),
represents the maximum search volume of the baseline word relative to the combined term
(obtained by comparing maximums in the time series data). The error term, 𝜖𝑠,𝑤 , is composed of
two parts: 𝑉𝑠,𝑐&𝑤 , the search volume for searches that contain both the baseline word and the
word of interest and, 𝜌, random error introduced by Google Trends’ use of sampling to produce
the search volumes. The 𝑉𝑠,𝑐&𝑤 component of the error term can be minimized by choosing a
baseline term that is unlikely to be included in a search with the term of interest. Finally, to
reduce any potential error caused by sampling or other anomalies, for each the set of estimated
search volumes 𝑉𝑠,𝑤 (s=1,2…j), factors (𝒇) are extracted
(4) 𝐸(𝑉𝑠,𝑤 |𝒇) = 𝛼𝑖0 + 𝛼𝑖1 𝑓1 + ⋯ + 𝛼𝑖𝑘 𝑓𝑘
Such that
(5) 𝐶𝑜𝑣(𝑉𝑠,𝑤 , 𝑉𝑗,𝑤 |𝒇) = 0
for all values of 𝑠 where 𝑠 ≠ 𝑗 (Bartholomew, Steele, Moustaki and Galbraith 2008). The primary
factor score is taken as the estimated subregional search volumes for the term of interest.
Illustration
To illustrate this method, estimated state-level search volumes are produced for the terms
“immigrants” and “illegals.” The baseline term used to produce the estimates is “noodle.” This
term is employed because its usage is relatively consistent over time and it narrowly meets the
40
minimum search volume to produce observed estimates in each state. It is desirable to use words
that are close to, but narrowly above, the absolute minimum threshold for each state in order to
avoid rounding error in the estimates for the combined searches (all search volumes are reported
by Google Trends as whole numbers between 0 and 100).
[TABLE 2.1 ABOUT HERE]
Table 2.1 shows columns for the observed values for “immigrants” (notice 5 “N/A”
values for states with absolute thresholds too low to produce observed values), observed values
for the baseline term, “noodle,” and observed values for combined term “noodle” and
“immigrants.”4 The final column shows the estimated values for this sample produced by
subtracting the baseline values multiplied by the maximum relative value for the baseline term
(in this case 55) over 100 (by definition the maximum for the combined term) from the values
for the combined term. Note that the observed values for “immigrants” are included in the table
only for reference and are not used in the calculations. Although, the scale for the estimated
values are different, the estimated values correlate with the observed values at .97. This process
is repeated three times and factors are extracted from the three sets of observed values using
exploratory factor analysis. The eigenvalues produced by exploratory factor analysis strongly
indicate that a one factor model is preferred and this single factor explains nearly all of the
variance in the observed variables. The factor extracted from the three sets of estimated values
for “immigrants” correlates with the observed values at 0.98 and the relationship is shown in
figure 2.1.
[FIGURE 2.1 ABOUT HERE]
4
Note that estimates for the District of Columbia are computed, though these values are not used in the
subsequent statistical analysis.
41
Figure 1 clearly shows a strong relationship between the observed values and estimated values,
indicating that the estimated values are a valid measure of state-level search volume for the term
“immigrants.” This procedure is repeated for the term “illegals,” another termed used in chapter
5, and a figure showing the factor extracted from the three samples is shown below plotted
against observed values in figure 2.2.
[FIGURE 2.2 ABOUT THERE]
Although there are far fewer observed values available for comparison, a very strong relationship
is evident in the plot between the observed values and the estimated values, again indicating that
the estimated values for the term “illegals” are a valid measure of the search volume for that
term. The state-level search volumes for the remaining terms used in the analysis in chapter 5,
“guns” and “jobs,” are observed directly as they are searched frequently enough that the
minimum threshold is obtained in all 50 states.
Tea Party Event Data
The Tea Party Patriots, the largest and most grassroots of all Tea Party organizations, maintains a
website that, at the time of data collection, contained an event calendar which allowed local
chapters to upload information about the time and location of their events.5 These events include
activities such as meetings, talks and protests. Using an automated script, I scraped this event
calendar and constructed a data set of all Tea Party events uploaded to the calendar between the
dates of January 1, 2009 and August 31, 20116. Over the course of this time period, 6,193 events
were uploaded to the event calendar by local groups from across the country. Of these 6,193
local events, I was able to match 5,851 of these events to counties and states using a dataset of all
5
6
Data come from http://www.teapartypatriots.org/events/. Collection dates were 6/2011-8/2011
The script used to obtain this data is available in the appendix at the end of this chapter.
42
named locations in the United States from the United States Geological Survey. Some events
were not able to be matched because the name was too vague, encompassed too wide of an area
or was inscrutable. Given that the information included in each event record included the date of
the event, it is possible to analyze this data in both longitudinal and cross sectional contexts.
Figure 1.2 from the previous chapter shows the events plotted over time with a second
measure of Tea Party activity, nationwide aggregate search frequency for the term “Tea Party,”
plotted alongside Tea Party events. The Tea Party event trends are consistent with both the
historical record of the movement as well as with the measure of Google searches for the term
Tea Party. This indicates that the Tea Party event measure seems to have a good level of face
validity and consistency with other measures. Table 2.2, below, shows the frequency of Tea
Party events over the full time period that occurred in each state.
[TABLE 2.2 ABOUT HERE]
As is evident in Table 2, there is a great deal of variability in the distribution of Tea Party Patriot
events within the United States. While, the Tea Party event measures seems to have validity with
respect to the national event count over time, it is also important to verify the validity of these
measure with respect to variability across geographic regions. A possible alternative measure of
Tea Party activity across the country is the state-level Google search frequency during the time
frame covered by the data relating to the Tea Party movement. Figure 1 shows state-level Google
search frequency for the term “Tea Party” plotted against the number of events per 100,000
residents that took place in each state7. Both metrics cover the time period from January 1, 2009
to August 31, 2011.
[FIGURE 2.3 ABOUT HERE]
7
The word “Boston” is excluded from this search term to filter out searches for the Boston Tea Party.
43
The correlation between state-level searches for the term “Tea Party” and the per-capita Tea
Party event count in each state is .43 (p<.01). Although this relationship is of moderate strength
it does suggest that the two measures are tapping the same underlying variable – local Tea Party
activity. One reason that the state-level search activity may not correlate more strongly with
state-level events is that, rather than representing local interest in Tea Party participation, the
search activity might be, to some extent, a response to national media coverage of Tea Party
activity. This might explain why Tea Party searches track more closely to national time-series
variation in Tea Party activity than state-level variation.
One solution to this problem would be to use a term that more exclusively focuses on
searches for information about local Tea Party events. Unfortunately, joint terms such as “Tea
Party” and ”events” have search frequencies that are too low to collect adequate state-level data
even using the techniques described in the above section of this chapter. Better confirming the
state-level validity of the Tea Party event data will, therefore, continue to be an important goal in
future research.
CONCLUSION
This chapter outlines some of the novel methods and data sources employed in the following
chapters. First, this chapter discusses the existing literature and research on using internet data
sources produced by the activity of online users in the course of normal use, such as crafting
messages on social media platforms or searching for terms of interest on search engines. The
existing literature has demonstrated that the data produced by this activity can be useful in
constructing metrics of public attention and public discourse. The chapter goes on to propose a
44
method for obtaining measures of public interest in selected topics using state-level data on
aggregate Google search frequencies.
Finally, this chapter discusses data on Tea Party Patriot events that were obtained by
scraping the group’s online event calendar. While, on the aggregate, national-level, this data
seems to have face validity, more work is necessary in order to confirm the validity of the
measure in terms of variation in events between states.
45
Table 2.1: Tea Party Patriots events per state
State
Alabama
Events
% of total
Events/
100k pop
State
Events
% of total
Events/
100k pop
110
1.88
2.47
Montana
52
0.89
5.76
Alaska
23
0.39
3.67
Nebraska
57
0.97
3.33
Arizona
109
1.86
2.12
35
0.60
1.75
Arkansas
83
1.42
3.10
Nevada
New
Hampshire
27
0.46
2.18
California
845
14.44
2.49
New Jersey
115
1.97
1.37
Colorado
164
2.80
3.81
New Mexico
80
1.37
4.40
Connecticut
118
2.02
3.46
New York
218
3.73
1.15
227
3.88
2.82
0
0.00
0.00
201
3.44
1.77
44
0.75
1.28
Delaware
25
0.43
3.19
North Carolina
Florida
377
6.44
2.36
North Dakota
Georgia
354
6.05
4.32
Ohio
6
0.10
0.50
Oklahoma
Idaho
16
0.27
1.24
Oregon
40
0.68
1.17
Illinois
216
3.69
1.74
Pennsylvania
315
5.38
2.56
Indiana
216
3.69
3.55
Rhode Island
6
0.10
0.57
Iowa
57
0.97
1.95
South Carolina
68
1.16
1.69
Kansas
71
1.21
2.64
South Dakota
Kentucky
81
1.38
2.00
Tennessee
Louisiana
66
1.13
1.48
Maine
14
0.24
Maryland
55
Massachusetts
92
Michigan
Minnesota
Hawaii
Mississippi
Missouri
0
0.00
0.00
98
1.67
1.72
Texas
342
5.85
1.64
1.10
Utah
26
0.44
1.16
0.94
1.04
Vermont
12
0.21
1.97
1.57
1.45
Virginia
95
1.62
1.34
139
2.38
1.40
Washington
172
2.94
2.92
75
1.28
1.52
West Virginia
52
0.89
2.88
20
0.34
0.70
Wisconsin
88
1.50
1.64
127
2.17
2.27
Wyoming
22
0.38
4.46
46
Table 2.2: Observed values for “immigrants,” “noodle,” “immigrants” OR “noodle” and the
estimated “immigrants” values
Subregion
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Immigrants
55
N/A
77
56
73
72
87
79
85
60
65
98
71
81
72
87
66
71
51
74
74
77
66
85
52
60
N/A
69
62
61
73
54
100
69
N/A
65
51
60
75
82
71
61
59
70
62
79
54
77
N/A
80
N/A
Noodle
44
64
53
51
71
78
76
57
71
49
60
84
57
100
95
93
66
63
52
64
77
77
68
82
40
69
97
90
51
64
73
51
89
60
64
81
57
79
80
53
55
70
51
60
66
67
53
91
51
83
71
Immigrants | Noodle
52
64
69
58
79
82
87
71
82
58
68
98
68
100
94
98
72
72
56
73
82
83
73
89
48
71
86
88
60
68
79
56
99
69
68
81
58
79
84
71
67
71
59
70
70
79
58
92
62
89
65
47
Est. Immigrants
27.8
28.8
39.85
29.95
39.95
39.1
45.2
39.65
42.95
31.05
35
51.8
36.65
45
41.75
46.85
35.7
37.35
27.4
37.8
39.65
40.65
35.6
43.9
26
33.05
32.65
38.5
31.95
32.8
38.85
27.95
50.05
36
32.8
36.45
26.65
35.55
40
41.85
36.75
32.5
30.95
37
33.7
42.15
28.85
41.95
33.95
43.35
25.95
Figure 2.1: Observed Immigrants vs Estimated Factor
48
Figure 2.2: Observed Illegals vs Estimated Factor
49
Figure 2.3: Tea Party Search plotted against events
50
Appendix 2:
Source code for script used to scrape Tea Party event calendar:
#importing modules
import BeautifulSoup
from BeautifulSoup import BeautifulSoup
from BeautifulSoup import BeautifulStoneSoup
import re
import urllib2
import urlparse
##################
####NOTE: This program rescrapes the tea party websites to update the data collected
earlier
#### Updated: September 6th 2011
#### Run:September 6th 2011
##################
###PART I: THE LOCAL CHAPTERS#####
tea_base="http://www.teapartypatriots.org/state/"
state = ["Alabama", "Alaska", "Arizona", "Arkansas", "California", "Colorado",
"Connecticut", "Delaware", "Florida", "Georgia", "Hawaii", "Idaho", "Illinois",
"Indiana", "Iowa", "Kansas", "Kentucky", "Louisiana", "Maine", "Maryland",
"Massachusetts", "Michigan", "Minnesota", "Mississippi", "Missouri", "Montana",
"Nebraska", "Nevada", "New_Hampshire", "New_Jersey","New_Mexico", "New_York",
"North_Carolina", "North_Dakota", "Ohio", "Oklahoma", "Oregon", "Pennsylvania",
"Rhode_Island", "South_Carolina","South_Dakota", "Tennessee", "Texas", "Utah",
"Vermont", "Virginia", "Washington", "West_Virginia", "Wisconsin", "Wyoming"]
pagel=[]
#this is the page list
#link_list = [] #all links found will be stored here
link_count = []
for i in range(0,len(state)):
tea_add=tea_base+state[i]
page=urllib2.urlopen(tea_add)
soup = BeautifulSoup(page)
link_list=soup.findAll('a')
link_count.append(0)
for a in link_list:
if str(a).find("GroupNew")>0:
link_count[i]=link_count[i]+1
print state[i], ":", link_count[i]
#print soup.prettify()
#state_key = {}
#linking state names to link counts
#c= 0
#for s in state:
#
state_key[s]=link_list[c]
#
c=c+1
#c=0
#for s in state_key.keys():
#
print s, ":", link_list[c]
#
c=c+1
51
######PART II: THE CALENDAR#######
#############
### Rewritten: September 6th, 2011
### Rerun:
##############
# Note: this state vector is different than the one in Part I (no underscores)
state = ["Alabama", "Alaska", "Arizona", "Arkansas", "California", "Colorado",
"Connecticut", "Delaware", "Florida", "Georgia", "Hawaii", "Idaho", "Illinois",
"Indiana", "Iowa", "Kansas", "Kentucky", "Louisiana", "Maine", "Maryland",
"Massachusetts", "Michigan", "Minnesota", "Mississippi", "Missouri", "Montana",
"Nebraska", "Nevada", "New Hampshire", "New Jersey", "New Mexico", "New York", "North
Carolina", "North Dakota", "Ohio", "Oklahoma", "Oregon", "Pennsylvania", "Rhode
Island", "South Carolina", "South_Dakota", "Tennessee", "Texas", "Utah", "Vermont",
"Virginia", "Washington", "West Virginia", "Wisconsin", "Wyoming"]
cal_base="http://www.teapartypatriots.org/Today.aspx?date="
#pagel = []
#mn_tot= []
#yr_keep=[]
locst=[] #state event takes place in
lochp=[] #local chapter name
city=[]
# city name
c=0
tester=1
mlen = [31, 28, 31, 30, 31,30, 31, 31, 30, 31, 30, 31]
mpy=13
for yr in range(2011,2012):
#loop through years
for mn in range(3,9):
#loop through months
#link_list2 = []
#no longer used
for d in range(1,mlen[mn-1]+1):
#loop through days
cal_add=cal_base+str(mn)+"/"+str(d)+"/"+str(yr)
page=urllib2.urlopen(cal_add)
soup = BeautifulSoup(page)
bees=soup.findAll('b')
bv_place=0
#This keeps track of the location in bees
for bs in bees:
#print bs
for s in state:
if str(bs)=="<b>"+s+"</b>":
locst.append(s)
lochp.append(bees[bv_place+1])
#print c,":",locst[c],":",lochp[c],":",yr,":",mn,":",d
bv_place=bv_place+1
dees=soup.findAll('div')
for ds in range(0,len(dees)):
if len(dees[ds])>1:
dsx=str(dees[ds].contents[1])
if dsx.find("<b>")==0:
if dsx.strip("<b>").strip("</") in state:
city.append(dees[ds].contents[3])
print
c,":",locst[c],":",city[c].strip("u'\r\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t"),":",lochp[c],
":",yr,":",mn,":",d
c=c+1
52
Chapter 3: Individual-Level Predictors of Tea Party Support and Activism
This chapter seeks to use individual level data to investigate the characteristics of Tea Party
activists and supporters as well as to test how existing theories of right-wing mobilization map
onto the Tea Party Movement at the individual level. Specifically, this chapter will assess the
demographics of Tea Party activists and the extent to which they perceive threats to their cultural
and economic status, as measured by polling data, from a number of sources including,
immigration, racial diversity, gay rights, unemployment, downward economic mobility and
taxation. Each of these issues is a potential source of cultural or economic threat – which have
been identified by previous literature as sources of right-wing mobilization and which there is
reason to suspect may underlie support for and activism in the Tea Party Movement.
The analysis in this chapter will use data from a New York Times and CBS News poll
with an oversampling of Tea Party activists and supporters to test the effect of these variables on
both participation in and support for the Tea Party Movement. Specifically, a multinomial
logistic regression model is estimated predicting individual-level tea party activism, tea party
support and non-support. The findings indicate that both Tea Party supporters and activists are
more conservative and express higher levels of discomfort with racial minorities and immigrants
than non-supporters. However, Tea Party activists, those who actually participate in Tea Party
events, are differentiated from supporters by much higher levels of frustration with taxes and
much greater fear of downward economic mobility. These findings indicate that, at the time of
the survey, a sense of economic threat was an important factor in driving participation in the Tea
Party, rather than simple support.
53
Finally, using polling data, this chapter will analyze how individual-level support for the
Tea Party has changed over the course of the movement. Previous work (e.g. Williamson et al
2011) shows that support for the Tea Party movement shrank over time after the early waves of
mobilization and publicity in 2009. Over the course of the movement, the Tea Party also faced
and responded to a number of different policies and changes in the political opportunity structure
such as the passage of the Affordable Care Act and the opportunity to engage with the 2010
midterm congressional elections. As previous research has indicated (e.g. Heaney and Rojas
2011), structural and political changes tend to alter the demographics of the supporters and
activists in political movements. Given these changes, it is important to understand how the
composition of the Tea Party movement evolved over the course of time.
The CBS News/New York Times poll from May 2010 is the only publicly available poll
that I am aware of that contains a substantial oversampling of Tea Party activists – most other
polling simply gauges support for the movement. This makes it impossible to use individual data
to study changes in the characteristics of Tea Party activists over time. However, using other
polling data, it is possible to examine changes in the characteristics of Tea Party supporters over
time. The analysis in this chapter uses for surveys from the PEW organization from February
2010 (the organization’s earliest survey asking about Tea Party support), June 2010, January
2011 and June 2011. The findings of this analysis indicate that the Tea Party’s base of support
became more conservative and more partisan in its support for the Republican Party over time.
Additionally, Tea Party’s base of support became whiter later in the movement. Other
characteristics, such as identification with evangelical Christianity, gender and income remain
unchanged. These findings suggest that, on the individual level, the Tea Party movement became
54
more highly integrated and aligned with the Republican Party and conservative politics in the
later stages of the movement.
PREVIOUS LITERATURE AND MOTIVATION
Right-wing Movements as a Reaction to Threat
Previous literature on right-wing and conservative movements has viewed mobilization primarily
as instances of high-status groups mobilizing to protect their status against perceived threats to
their cultural, economic or political positions (e.g. Gusfield 1963; McVeigh 1999). For example,
Gusfield argues that the temperance movement in the United States was motivated by a sense
among protestants that the increasing social position and cultural power of recent waves of
Catholic immigrants were a threat to their position as the dominant cultural group in American
society. As such Protestants mobilized to enact their social values into law, thereby reaffirming
their cultural status. The prohibition against alcohol, according to Gusfield, simply served as a
medium through which Protestants could mobilize to assert their cultural dominance. On other
hand, McVeigh (1999) argues that Klan mobilization in the United States in the 1920s was a
reaction by wealthy groups of rural landowners and professionals who saw the growing power of
urban manufacturing as a threat to their economic standing, and their supply of cheap labor, and
who saw urban political machines as a threat to their political standing. Race, religion and
ethnicity, according to McVeigh, served primarily as a symbol around which the movement
could unite. Similarly, Van Dyke and Soule (2002) argue that the rise of Patriot Militia groups in
the 1990s were a response to the declining economic prospects of rural whites.
Who Supports the Tea Party and Why
55
There is reason to believe that, like earlier right-wing and conservative movements, Tea
Party activists and supporters are reacting to real or perceived threats to their status. However,
the particular forms that threat motivating Tea Party activism are not immediately clear from a
superficial examination of the movement. Most Tea Party groups and their prominent supporters
claim the movement is motivated by a desire to reduce government spending, taxation and
regulation which they view as threatening to their economic livelihood (Armey 2009; “Tea Party
Patriots,” n.d). Other work, including journalistic accounts (e.g. Zernike 2010) have pointed
toward the stress of job loss and potential downward mobility as a factor in motivating Tea Party
activism. In this sense, Tea Partiers could be understood as fearing the effects of taxation or
regulation on their economic prosperity or as simply blaming their unemployment on the policies
or politicians targeted by the movement.
At the same time, the Tea Party is largely socially conservative and many observers have
argued that hostility toward racial, ethnic and religious minority groups motivates a great deal of
Tea Party support and activity (e.g. Potok 2010; Abramowitz 2011; Williamson et al 2011).
Williamson et al (2011) have argued that the Tea Party is largely motivated by xenophobia and
the threat of cultural, racial and ethnic change in the U.S. – they argue that the election of
Obama, a man who they view as alien to their way of life, was the initial catalyst for the
emergence of the movement. In fact, they argue, consistent with others like Gilens (1999), that
the Tea Party’s opposition to government spending is primarily motivated by the fact that they
view this spending as benefiting non-deserving groups. The boundaries of these non-deserving
groups are largely defined by race, culture and ethnicity. Through their qualitative interviews
with Tea Party activists in the Boston area, they identify illegal immigrants as one of the main
targets of the Tea Party’s hostility. The Tea Party members’ hostility toward illegal immigrants
56
was based primarily on fears that such immigrants would unfairly receive benefits paid for
through taxes – taxes which the Tea Party activists viewed themselves as disproportionately
paying despite the fact that many of those interviewed were, themselves, unemployed.
Using survey data to systematically understand the demographics of the Tea Party
movement as well as the forms of threat that are salient to Tea Partiers at the individual level is
important, however, it is also important to systematically examine how Tea Party activists, those
who take part in Tea Party events or donate money, compare to Tea Party supporters, those who
simply express support for the movement in surveys. Previous literature has shown that the
processes that mobilize individuals to protest or participate in social movements are not the same
as those that are simply associated with support (e.g. Snow, Zurcher and Olsen 1980) while other
research has shown that different levels of engagement and activism are associated with different
individual-level characteristics (McAdam 1986; DiGrazia 2014). This research essentially finds
that various social constraints can prevent individuals who support the cause from participating
in activism. For example, those who have fewer constrains on their time and who have fewer
other responsibilities (e.g. children to care for) are more likely to assume to costs of participating
in activism. Additionally, those who engage more heavily tend to be more ideologically extreme
than those who participate at more moderate levels. As Tea Party activists constitute a small
minority of Tea Party supporters, it is important to understand how activists differ from those
who simply support the movement, especially as much of the existing public and academic
commentary on the Tea Party that has used survey and polling data has focused on supporters.
While some, such as Williamson et al. (2011), have studied Tea Party activists, they have not
been systematically compared to Tea Party supporters.
57
Changes in Movements Over Time
As has been noted in previous research (e.g. Williamson, Skocpol and Coggin 2011) and
demonstrated again here (see Figure 3.3), the Tea Party grew less popular with the public after
the spring of 2010. Given that the group of people supporting the Tea Party was shrinking over
this time period, it is likely that the demographics of the movement shifted systematically in
response to changes in the national political environment and the political opportunity structure.
Previous research on social movements shows that activists and supporters can be mobilized or
demobilized in response to changes in the political environment. Additionally, the boundaries,
goals and targets of a movement can change under these circumstances (Tarrow 1993;
Staggenborg 1998; Fligstein and McAdam 2011). For example, Heaney and Rojas (2011) show
that anti-war protesters affiliated with the Democratic Party were demobilized after the election
of Barack Obama, as his election caused them to perceive the war as being less threatening. After
Obama’s election, the anti-war movement became smaller and heavily dominated by those with
no affiliation to either major political party. Like the anti-war movement, the Tea Party
movement faced considerable changes in its political environment over the course of the
movement. Figure 3.1 below shows Google Trends data showing search volumes for joint
searches (searches that include both words) on the words “Tea Party” and “Health” and “Tea
Party” and “Election.”
[FIGURE 3.1 ABOUT HERE]
Figure 3.1 shows that during the period in which the health care reform debate was dominating
national discourse, joint searches between the Tea Party and health care peaked while joint
searches between “Tea Party” and “election” peaked during the 2010 election cycle. This
provides some evidence indicating that the Tea Party movement shifted its focus from the health
58
care reform debate to the congressional election. As such, one plausible hypothesis is that the
Tea Party movement, defined as the group of people supporting the movement, became more
partisan and Republican as it began to focus directly on party politics, while a wider variety of
individuals may have expressed support for the movement in the earlier stages when the focus
was on health care reform.
In addition to analyzing the sources of threat associated with Tea Party support and
activism, this chapter will also provide an analysis of how the demographics of the movement’s
base of support changed over time.
METHODS AND DATA
Data Sources
The data for the analysis in this chapter comes from several sources: an April 2010 NYT/CBS
News poll, as well as several PEW political surveys collected between February 2010 and June
2011. The NYT/CBS News poll has 1,591 observations (reduced to 1308 after listwise deletion
on missing observations) and is national in scope and constitutes a sample of the adult population
collected using both cell phones and landline numbers. The survey oversamples Tea Party
activists with 194 individuals who have actively participated in Tea Party activities. The PEW
surveys, for the months of February 2010, June 2010, January 2011 and June 2011, contain
1,383, 1,802, 1,503, and 1,502 respondents, respectively. Each poll was collected as a random
digit dialing telephone survey of the adult US population. February 2010 is the earliest point at
which PEW began polling on attitudes toward the Tea Party movement. Most other polling
houses also began polling on this item in early 2010.
59
NYT/CBS News Measures
The dependent variable used in this analysis consists of a categorical item indicating whether an
individual is a Tea Party activist, a Tea Party supporter or a non-supporter. Activism can include
attending protests or meetings or donating money, supporters are those who indicate support for
the movement but have not actively participated and non-supporters are those who neither
indicate support nor take part in activities. The three categories are mutually exclusive.
The independent variables used in the individual-level analysis consist of many standard
measures of demographic characteristics as well as items measuring attitudes toward the
economy and cultural issues. These economic and cultural items are intended to measure a sense
of economic or cultural threat.
Basic demographic measures
Income is measured in six categories: under 15 thousand, 15 to 30 thousand, 30 to 50 thousand,
50 to 75 thousand, 75 to 100 thousand and over 100 thousand per year. White consists of a
dummy variable measuring whether the respondent identifies as white or a member of another
racial group. Gender is measured as a dummy variable coded as one if the respondent is female,
and zero if the respondent is male. Education is measured in five categories: less than high
school, high school graduate, some college, college graduate, and post-graduate. Self-described
social class is measured in five categories: upper, upper-middle, middle, working and lower. The
measure of political ideology is measured is a five point scale: very liberal, somewhat liberal,
moderate, somewhat conservative, and very conservative. Also included are dummy variables for
religious identities, including Catholic and Protestant, with others as the reference category and
for Democrat and Republican with independents as the reference category.
60
Measures of Cultural Attitudes and Cultural Threat
Measures of cultural attitudes include items measuring attitudes toward legal immigration, illegal
immigration, gay marriage, abortion, gun control and an item asking respondents if they believe
“too much as been made of the problems facing black people.” The immigration items consist of
a dummy variable indicating whether or not the respondent thinks illegal immigration is a serious
problem as well as a dummy variable indicating whether or not the respondent wants to decrease
foreign immigration to the US. These items are intended to measure hostility toward immigrants
as an outgroup. The gay marriage measure is a dummy variable indicating whether or not the
respondent supports the legalization of gay marriage. Similarly, the abortion item is a dummy
variable indicating whether or not the respondent opposes legal abortion. The items measuring
attitudes toward abortion and gay marriage are intended to measure attitudes on traditional
‘culture war (e.g. Luker 1985) issues that animated conservative politics in the late 20th century.
The gun control measure is also a dummy variable measuring whether or not the respondent
supports increased gun control legislation. The item measuring gun control is included as
research finds that opposition to gun control in the United States is increasingly associated with
conservative politics and is associated with hostility toward and fear of racial minorities
(O’Brien, Forrest, Lynott and Daly 2013). Finally, an item is included that is intended to measure
anti-black sentiment. This item consists of a dummy variable indicating whether or not
respondents indicated agreement with the statement “too much has been made of the problems
facing blacks.”
Measures of Economic Threat
61
Other items measuring attitudes toward the economy include a dummy variable
measuring whether the respondent thinks his or her taxes are fair and a dummy variable
measuring whether the respondent believes he or she will face a decline in economic status in the
near future. Additionally, the measures of income and self-identified social class described above
in the basic demographic measures section should help to further establish economic patterns in
Tea Party support and activism. An item measuring whether the respondent is unemployed is
also included in all models as a measure of economic stress. Descriptive statistics for all
measures in the CBS/NYT data are shown below in Table 3.1.
[TABLE 3.1 ABOUT HERE]
Pew Measures
The second set of individual surveys used in this chapter’s analysis are PEW political surveys
collected between February 2010 and June 2011. The time frame of this analysis begins in
February 2010 as this was the earliest survey fielded by the PEW organization that polled
respondents on their attitudes on the Tea Party Movement (other polling organizations began
polling on this topic at a similar point in time). This time frame was selected because it allows an
examination of the Tea Party movement through the 2010 elections and the period beyond and it
is consistent with the time period employed in the analysis of event data in chapters 4 and 5.
Each survey is weighted to eliminate potential bias introduced by the inclusion of cell phones
and to match sample demographic estimates to population parameters.
Dependent variable
62
The dependent variable used in the analysis of the PEW data is a binary measure indicating
whether the respondent supports or does not support the Tea Party Movement. Three different
question wordings were used in the surveys. The February 2010 survey asked respondents to
report whether or not they view the Tea Party movement very favorably, somewhat favorably,
somewhat unfavorably or very unfavorably. This item is collapsed into a binary form indicating
a favorable or unfavorable view. The June 2010 and January 2011 surveys use an item asking
whether respondents agree strongly, agree somewhat, disagree somewhat or disagree strongly
with the Tea Party Movement. This item, similarly, is collapsed into a binary variable indicating
agreement or disagreement. Finally, the June 2011 survey uses a binary item measuring
agreement or lack of agreement1.
Independent Variables
Independent variables used in the analysis of the PEW data include measures of basic
demographic information including sex, age, race and income. Age is measured in years, ranging
from 18 to 97. Income is measured in nine in categories: less than 10 thousand, 10 to 20
thousand, 20 to 30 thousand, 30 to 40 thousand, 40 to 50 thousand, 50 to 75 thousand, 70 to 100
thousand, 100 to 150 thousand and over 150 thousand. Sex and race are dichotomous, with sex
coded as 1 for female and 0 for male and race coded as 1 for white or 0 for nonwhite. Other
independent variables include a binary item measuring whether or not the respondent considers
him or herself a born again Christian, and a binary item measuring whether or not the respondent
identifies as a republican. Additionally, an item is included measuring education in 7 discrete
categories: elementary, less than high school, technical, some college, college graduate and post-
1
Robustness tests with dummy variables for question type indicate that question wording does not significantly
affect the outcome of the analysis.
63
graduate. Finally a measure of political ideology is included with five categories: very
conservative, conservative, moderate, liberal and very liberal. Additional control variables (not
shown in the tables) include an item measuring whether or not the call was to a landline or cell
phone and a dummy for variable indicating which version of the dependent variable question is
used. Descriptive statistics for independent variables in the pooled sample are shown below in
Table 3.2.
[TABLE 3.2 ABOUT HERE]
Analytical Strategy
The analysis of the NYT/CBS data uses multinomial logistic regression to examine the
relationships between the independent variables and participation in and support for the Tea
Party movement relative to non-support. The multinomial logistic regression model is an
extension of binary logistic regression that simultaneously estimates all combinations of the
response categories of a nominal variable using binary logistic regression (Long 1997). This
modeling strategy allows for the analysis of differences between non-supporters and supporters,
non-supporters and activists and supporters and activists.2 Two models are estimated: one
including only a basic set of demographic measures as independent variables and the other
including all independent variables. Comparisons between non-supporters and supporters, nonsupporters and activists and supporters and activists are reported for both models.
In order to analyze the PEW data, I merge items from the four waves and estimate
logistic regression models on the pooled sample estimating the effects of the independent
2
To ensure proper model specification, a series of likelihood ratio tests were performed on the hypothesis
𝐻0 : 𝛽1,𝑎|𝑏 = ⋯ = 𝛽𝑘,𝑎|𝑏 = 0 for categories a and b for each possible combination of categories. In each case the
null hypothesis was rejected indicating that none of the response categories can be collapsed together (Long
1997).
64
variables and time on Tea Party support. I also estimate separate models interacting time with
each of the independent variables to test whether the effect of each variables changes
significantly over time.
RESULTS
Analysis of Tea Party Activism, Support and Non-Support
Figure 3.2 shows the basic demographic characteristics of Tea Party activists, Supporters and
non-supporters. Each horizontal bar represents the proportion of Tea Party activists, supporters
or non-supporters who are 60 or older, Catholic, Protestant, College Educated, Female, White or
earn $50,000 or more. The figure shows that Tea Party supporters and activists have a similar
profile and tend to have some important differences from the population of non-supporters.
Specifically, Tea Party activists and supporters tend to be older, more educated, more protestant
and more male than non-supporters. Activists and supporters also tend to be whiter and have
higher incomes than the non-supporting population. Taken together Figure 2 shows Tea Party
activists and supporters to occupy relatively privileged positions in society: older, whiter, more
male, better educated and wealthier. However, as is evident in Table 3.3, many of these
characteristics do not define participants and supporters in the Tea Party movement after
controlling for other characteristics such as partisanship, political ideology and attitudes– in
other words, Tea Partiers may be whiter, more male, more protestant and wealthier than the
general population, but no more so than conservative republicans in general.
[FIGURE 3.2 ABOUT HERE]
[TABLE 3.3 ABOUT HERE]
65
The results of the multinomial logistic regression analysis are shown in table 3.3. The
results in columns one and two (non-supports to supporters) show the effect of each independent
variable on the difference between Tea Party supporters and non-supporters.
The results, in the base model, indicate that income, age, education, being white and
being protestant have a positive effect on the odds of being a supporter vs a non-supporter, while
being female has a negative effect. However, many of these results become insignificant in the
full model with the inclusion of measures of ideology, partisanship and threat. In the full model,
age has a significant, positive effect on the probability of being a supporter over a non-supporter.
Specifically, each additional year of age is associated with a factor increase in the odds of being
a supporter equal to 1.033 (p<.001). Additionally, each education category is associated with an
increase in the odds of being a supporter relative to a non-supporter equal to a factor of 1.603
(p<.001). The effect of income, being white, being female and being protestant drop out of
significance in the full model, indicating that after controlling for measures of ideology,
partisanship and threat, these factors do not significantly distinguish supporters from nonsupporters.
Tea Party supporters are both more conservative and more republican than nonsupporters net of other factors. Each one point increase on the conservative ideology scale is
associated with a factor increase in the odds of being a Tea Party supporter relative to an activist
equal to 2.189 (p<.001). Additionally being a Democrat is associated with a factor decrease in
the odds of being a supporter equal to 0.255 (p<.001), net of other factors. These findings
indicate that Tea Party supporters are more conservative and less Democratic than others even
taking into account demographic characteristics and other social attitudes.
66
Several items measuring cultural or racial threat are also statistically significant in the full
model. Each one point increase on the scale measuring whether or not respondents believe that
illegal immigration is a problem is associated with a factor increase in the odds of being a Tea
Party supporter relative to a non-supporter equal to 1.427 (p<.05). Additionally, the item
measuring anti-black sentiment (whether respondents believe “too much has been made of
problems facing black people”) has a positive, significant effect. Those who answered in the
affirmative have odds of being a Tea Party supporter relative to a non-supporter that are higher
by a factor of 2.437 (p<.001). The item measuring support for gun control is also significant,
showing that those who support increased gun control have odds of being a supporter that are
lower by a factor of .347 (p<.001) compared to those who answered in the affirmative.
Among the items measuring economic threat, only the perceived risk of falling in social
class has a significant effect on the odds of being a supporter relative to being a non-supporter.
Believing that one is at risk of falling in social class is associated with a factor increase in the
odds of being a Tea Party supporter equal to 1.670 (p<.05) all else equal.
Columns 3 and 4 show the base model and full model for the comparison between nonsupporters and Tea Party activists. In many respects the results are similar to those for the
difference between supporters and non-supporters. The base model is substantially the same,
with positive, significant effects for income age, education, being white and being protestant and
a negative significant effect associated with being female. However, in the full model, income,
being white, being female and being protestant drop out of significance.
Additionally, in the full model, there is a positive and significant effect for conservative
ideology (p<.001) and a negative and significant effect associated with being a Democrat
(p<.05). Specifically, each one point increase on the conservative ideology scale is associated
67
with a factor increase in the odds of being a Tea Party activist relative to a non-activist equal to
3.776. Additionally, being a Democrat is associated with a factor decrease in the odds of being
an activist equal to 0.199.
Among the cultural threat items, each one point increase on the scale measuring the
extent to which respondents believe illegal immigration is a problem is associated with a factor
increase equal to 2.303 (p<.05). Additionally, answering affirmatively to the anti-black sentiment
item is associated with a factor increase in the odds of being a Tea Party activist relative to a
non-supporter equal to 2.438. Supporting gun control is associated with a factor decrease in the
odds of being an activist relative to a supporter equal to 0.159 (p<.001).
Finally, among the economic threat items, believing that one’s taxes are fair is associated
with a factor decrease in the probability of being a Tea Party activist associated with 0.450
(p<.01). Finally, believing that one is at risk of falling in social class is associated with a factor
increase in the odds of being a Tea Party activist relative to a non-supporter equal to 2.984
(p<.001).
Columns 5 and 6 show the results for the comparison between supporters and activists for
both the base model and the full model. In the base model there is a positive effect of both age
and unemployment, indicating that Tea Party activists are significantly older and significantly
more likely to be unemployed than Tea Party supporters. It is interesting to note that
unemployment is significant in neither of the two previous comparisons. Unemployment has a
negative and insignificant effect on the difference between supporters and non-supporters and a
positive and significant effect on the difference between activists and non-supporters. In other
words, neither supporters nor activists are significantly more or less likely to be unemployed
than non-supporters, but activists are significantly more likely to be unemployed than supporters.
68
The effect of unemployment could potentially be interpreted both in terms of biographical
availability (i.e. the unemployed have more time to take part in activities) or in terms of
economic threat (i.e. the unemployed feel that their economic status is under threat and see the
Tea Party as a means to take action against what they perceive to be the source of this threat).
However, the fact that the difference drops out of significance in the full model, after controlling
for other economic threat measures suggests that it might be the later – that unemployment is
associated with a sense of economic threat.
In the full model there are only three items that significantly differentiate between
supporters and activists as both the effects of age and unemployment drop out of significance in
the full model. Conservative ideology has a positive and significant effect on the odds of being
an activist relative to a supporter. Specifically, each one point increase on the conservative
ideology scale is associated with a factor increase in the probability of being an activist relative
to a supporter equal to 1.725 (p<.001). Additionally, believing ones taxes are fair is associated
with a decrease in the odds of being a supporter relative to an activist. Specifically, believing
one’s taxes are fair is associated with a factor decrease in the odds of being an activist relative to
a supporter equal to 0.621 (p<.05). Additionally, believing that one is at risk of falling in social
class is associated with a factor increase in the odds of being an activist rather than a supporter
equal to 1.787 (p<.05). Taken together, these results indicate that Tea Party activists are
significantly more conservative than Tea Party supporters (though not significantly more
Republican or less Democratic) and Tea Party activists experience a greater level of economic
threat as measured by the perceived risk of falling in social class and the sense that taxes are fair.
Analysis of Tea Party Support Over Time
69
Figure 3.3 shows the proportion of Americans expressing support for the Tea Party
movement from early 2010 through the summer of 2011. Over this time period Tea Party support
drops from over 33% to under 20%. Additionally, figures 3.4a and 3.4b show changes in the
values of several demographic independent variables over time for both Tea Party supporters and
non-supporters. In many respects, these data show that Tea Party supporters have characteristics
similar to those they are shown to have in Figure 3.2. Tea Party supporters consistently have
higher incomes, are less likely to be female and are more likely to identify as born again
Christians over the observed time period. Across many of the other items, there is increasing
divergence between Tea Party supporters and non-supporters over this time period. This is
particularly, true with respect to political ideology, education, racial composition and age. Here
we see Tea Party supporters become more conservative, more highly educated, whiter and older
relative to non-supporters over the observed time period.
[FIGURE 3.3 ABOUT HERE]
[FIGURES 3.4a and 3.4b ABOUT HERE]
[TABLE 3.4 ABOUT HERE]
Table 3.4 shows the results of logistic regression analyses predicting Tea Party support in
a pooled sample containing data from February 2010, June 2010, January 2011 and June 2011.
The effects shown are exponentiated beta coefficients. The base model shows significant effects
for political ideology, income, female, being a born again Christian, political partisanship
education and age. There is also a negative, statistically significant effect of time, indicating that
Tea Party support declined over the observed time period, confirming the visible trend in figure
3.3. Each one point increase on the 5 point political ideology scale (higher values indicating that
the respondent is more liberal) is associated with a predicted factor decrease in the odds of being
70
a Tea Party supporter equal to 0.488 (p<.001), net of all other factors. Each category increase on
the income scale is associated with an increase in the odds of being a Tea Party supporter equal
to 1.080 (p<.001). Being female is associated with a decrease in the odds of being a Tea Party
supporter equal to 0.633 (p<.001) and identifying as a born again Christian is associated with an
increase in the odds of supporting the Tea Party equal to 1.200 (p<.05). Being Republican has a
massive, positive effect on Tea Party support, with a factor increase in the odds equal to 5.30
(p<.001). For each year of age there is a factor increase in the odds of Tea Party support equal to
1.009 (p<.001). It is interesting to note that race does not seem to have a significant effect, net of
other variables, in the model.
The remaining columns in Table 4 show models with significant time interaction effects;
those interacting white, liberalism, Republican and Age with time. The interaction term for white
and time is positive, indicating that the effect of White on the odds of Tea Party support grows
over the observed time period, holding other factors constant. Similarly, there is a negative,
significant effect for the interaction between liberalism and time (p<.01). This indicates that
during later times the negative effect of liberalism on the odds of Tea Party support become even
larger in magnitude – identifying with more liberal political ideologies decreases the odds of Tea
Party support more in later time points than earlier ones. The interaction of Republican and time
has a positive significant effect (p<.05), indicating that the positive effect of being a Republican
on the odds of Tea Party support increases over time. Finally, the interaction between age and
time is also significant and positive (p<.01). Together, the results indicate that as the Tea Pary’s
base of support grew smaller, the movement became more conservative, more republican, older
and whiter.
71
CONCLUSION
The findings of this chapter begin by providing insight into who Tea Partiers are and
what they believe. Tea Party members tend to be conservative, republicans, white, male,
educated, protestant and possess high incomes. In this sense, the Tea Party bears demographic
similarities to the Republican Party in general (see Appendix 1 at the end of this chapter for an
analysis of the demographics of the Republican Party). However, the Tea Party also holds
particular views regarding cultural and racial issues as well as economic issues that indicate that
this is a group that feels particular threat to their statuses in these areas. Tea Party supporters and
activists are more likely to express hostility toward immigrants and African Americans than nonTea Party members, even after controlling for party identification and political ideology.
Additionally, Tea Party members –and particularly activists- are more likely to express concern
for their economic standing, being more likely to believe that their taxes are unfair and that they
are at risk of falling in social class. Tea Party activists express a significantly greater sense of
economic threat than supporters, indicating that economic threat, at least at the time of this
analysis, was a factor particularly likely to mobilize people in support of the Tea Party. Table 3
also provides some evidence that Tea Party activists are more likely to be unemployed than nonactivists. Taken together these findings support previous theories of right-wing and conservative
mobilization which have found them to be a reaction by high-status groups against threats to
their cultural and economic status. It is also consistent with some previous work on the Tea Party
movement, which has found that anti-immigrant sentiment and racial threat motivated much Tea
Party activism and support.
However, this chapter finds that, at least for the time period during which the data was
collected, what distinguishes those who mobilize from those who are merely supportive is a
72
sense of economic threat and the potential for downward mobility. This threat may be manifested
in a racialized manner (e.g. racialized opposition to the welfare state and taxation), however the
results suggest that economic concerns are a major driver of Tea Party activism.
This chapter also shows that the Tea Party’s base of support both shrinks and changes in
composition over time. Unfortunately, because polling houses started polling on the Tea Party
relatively late and because they rarely asked about activism, it is only possible to look at support
(rather than activism) starting in early 2010 (shortly before the Affordable Care Act was signed
into law). However, over the course of this time period, this chapter shows that the Tea Party’s
supporters became more conservative, more republican, whiter and older. In this sense, it appears
that the Tea Party’s base becomes more partisan and narrower – limited more to traditionally
conservative social groups.
The next chapter will build on the findings of this chapter by examining Tea Party
mobilization across the United States over time. Looking at Tea Party events on the county and
state-levels the next chapter will examine how the factors associated with mobilization changed
over time and, more importantly, how local context interacts with the national media and policy
environment to produce local Tea Party activism.
73
Table 3.1: Descriptive Statistics for CBS/NYT Data
Mean (weighted
pop estimate)
SE
Description
Income
White
Female
Education
South
Catholic
Protestant
Unemployed
Ideology
3.603
0.784
0.494
2.838
0.306
0.221
0.520
0.138
3.208
0.087
N/A
N/A
0.063
N/A
N/A
N/A
N/A
0.052
Republican
Democrat
Decrease Imm
0.300
0.318
0.410
N/A
N/A
N/A
Illegal Imm Prob
3.437
0.049
Gay Marriage
Problems Facing
Blacks
No Abortion
Gun Control
Taxes Fair
Job Worry
Social Class
Econ Fall Risk
0.405
0.299
N/A
N/A
0.193
0.383
0.635
0.624
2.524
0.399
N/A
N/A
N/A
N/A
0.049
N/A
74
Income categories
White=1
Female=1
Education categories
South=1
Catholic=1
Protestant=1
Unemployed=1
Ideology (greater values are
conservative)
Republican=1
Democrat=1
Does R want to decrease legal
immigration? 1=yes
How big of a problem is illegal
immigration? Greater values
indicate a greater problem.
Supports Gay Marriage=1
R thinks “too much made of
problems facing blacks”
R wants to prohibit abortion
R favors gun control
R believes taxes are fair
R is worried about losing job
Self-Identified Social Class
R fears personal economic
decline
Table 3.2: Pooled descriptive statistics for PEW independent variables.
Variable
Obs
Mean
Std. Dev. Min
Max
White
6190
0.763
0.426
0
1
Liberalism
5892
2.730
0.972
1
5
Income
5403
5.133
2.399
1
9
Female
6190
0.529
0.499
0
1
Born Again
6190
0.322
0.467
0
1
Republican
6190
0.437
0.496
0
1
Education
6164
4.756
1.632
1
7
Age
6094
51.942
17.929
18
97
75
Table 3.3: Individual-level multinomial logistic regression results showing effect of independent
variables on support for and participation in the Tea Party Movement (shown as exponentiated
betas; N=1308).
non-sup
non-act
sup-act
Income
Age
White
Female
Education
South
Catholic
Protestant
Unemployed
1.164**
(2.704)
1.035***
(6.409)
2.816***
(4.071)
0.580***
(-3.653)
1.367***
(4.508)
0.945
(-0.354)
1.833*
(2.505)
2.389***
(4.328)
0.687
(-1.182)
Conservative Ideology
Republican
Democrat
Decrease Immigration
Illegal Imm Problem
Support Gay Marriage
Anti-Black Sentiment
Opposition to Abortion
Gun Control
Taxes Fair
Job Worry
Social Class
Fall Risk
Constant
0.002***
(-13.363)
1.145
(1.680)
1.033***
(3.949)
1.089
(0.293)
0.812
(-1.087)
1.603***
(4.998)
0.758
(-1.345)
1.251
(0.713)
1.025
(0.093)
0.831
(-0.439)
2.189***
(6.084)
1.168
(0.743)
0.225***
(-4.801)
0.906
(-0.505)
1.427*
(2.064)
0.598
(-1.862)
2.437***
(4.546)
0.803
(-0.935)
0.347***
(-4.123)
0.724
(-1.627)
0.83
(-0.811)
1.105
(0.653)
1.670*
(2.212)
0.000***
(-6.203)
1.326***
(3.505)
1.049***
(7.825)
2.335*
(2.270)
0.525**
(-3.025)
1.372***
(3.544)
1.123
(0.532)
1.69
(1.605)
2.814***
(3.597)
1.806
(1.509)
0.000***
(-13.642)
z-statistics in parentheses, * p<0.05, ** p<0.01, *** p<0.001
76
1.285*
(2.325)
1.050***
(4.904)
0.507
(-1.588)
0.746
(-1.070)
1.650***
(4.208)
0.835
(-0.668)
0.873
(-0.322)
0.866
(-0.408)
1.682
(0.943)
3.776***
(7.213)
0.831
(-0.665)
0.199*
(-2.556)
0.88
(-0.486)
2.303*
(2.372)
0.479
(-1.795)
2.438***
(3.382)
0.993
(-0.021)
0.159***
(-4.039)
0.450**
(-3.131)
1.057
(0.198)
0.815
(-1.014)
2.984***
(4.021)
0.000***
(-7.490)
1.139
(1.663)
1.014*
(2.295)
0.829
(-0.454)
0.905
(-0.482)
1.004
(0.043)
1.187
(0.823)
0.922
(-0.241)
1.178
(0.572)
2.630*
(2.339)
0.069***
(-4.220)
1.123
(1.186)
1.016
(1.836)
0.465
(-1.736)
0.919
(-0.359)
1.029
(0.299)
1.1
(0.420)
0.697
(-0.960)
0.845
(-0.542)
2.023
(1.453)
1.725***
(3.302)
0.711
(-1.452)
0.883
(-0.203)
0.971
(-0.132)
1.613
(1.418)
0.802
(-0.596)
1
(0.001)
1.237
(0.826)
0.459
(-1.765)
0.621*
(-2.227)
1.273
(1.009)
0.738
(-1.700)
1.787*
(2.419)
0.006**
(-2.946)
Table 3.4: Regression results showing regression of Tea Party support on independent variables
over time. Only significant interactions shown. (N=5,158).
Base
White×Time
Lib×Time
Rep×Time
Age×Time
White
1.159
0.846
1.154
1.154
1.168
(1.491)
(-1.085)
(1.448)
(1.445)
(1.571)
Liberalism
0.488***
0.488***
0.585***
0.488***
0.486***
(-14.560)
(-14.544)
(-6.879)
(-14.551)
(-14.622)
Income
1.080***
1.079***
1.082***
1.080***
1.082***
(4.355)
(4.295)
(4.420)
(4.322)
(4.418)
Female
0.633***
0.633***
0.633***
0.632***
0.635***
(-6.163)
(-6.171)
(-6.156)
(-6.202)
(-6.125)
Born Again
1.200*
1.197*
1.198*
1.197*
1.194*
(2.329)
(2.294)
(2.302)
(2.297)
(2.258)
Republican
5.230***
5.231***
5.278***
4.096***
5.232***
(19.742)
(19.688)
(19.783)
(11.129)
(19.739)
Education
1.066*
1.066*
1.065*
1.066*
1.067*
(2.465)
(2.477)
(2.443)
(2.494)
(2.506)
Age (years)
1.009***
1.009***
1.009***
1.009***
0.999
(3.720)
(3.737)
(3.750)
(3.718)
(-0.185)
Time
0.784***
0.642***
1.068
0.691***
0.557***
(-7.105)
(-5.233)
(0.591)
(-6.049)
(-5.325)
White×Time
1.272**
(2.600)
Liberalism×Time
0.878**
(-2.921)
Republican×Time
1.204*
(2.519)
Age×Time
1.007**
(3.287)
Constant
0.449**
0.583*
0.286***
0.529*
0.722
(-3.231)
(-2.028)
(-4.300)
(-2.489)
(-1.139)
*p<.05 **p<.01 ***p<.001
t-statistics in parentheses.
77
Figure 3.1: Normalized search volume for joint searches for the words “Tea Party” and
“Election” as well as “Tea Party” and “Health” between January 1, 2009 and August 31, 2011.
Data from Google Trends.
78
Figure 3.2: Proportion of activists, supporters and non-supporters in various demographic
groups.
79
Figure 3.3: Percent of the Population Supporting the Tea Party Over Time
80
Figure 3.4a: Independent Variable Trends for Supporters and Non-Supporters
81
Figure 3.4b: Independent Variable Trends for Supporters and Non-Supporters
82
Appendix 3: Comparison of basic demographic characteristics for Republicans and nonRepublicans. Data from NYT/CBS News April 2010 Poll.
Not
Variable
Republican
Republican p(r-nr=0)
Ideology
2.96
3.82
<.001
White
0.71
0.95
<.001
Female
0.5
0.49
0.72
Education
2.82
2.89
0.35
Protestant
0.47
0.62
<.001
Income
3.49
3.84
<.001
83
Chapter 4: Tea Party Mobilization and the National Political Discourse
Chapter 3 explored the individual-level characteristics of Tea Party supporters and activists. It
found that Tea Party supporters and activists generally occupy high-status positions in American
society – they are likely to be conservative, wealthy, white, male, and educated. At the same
time, the findings indicate that these are not the only characteristics that set Tea Party activists
and supporters apart from the rest of the population. Tea Partiers tend to express more hostility
toward racial and ethnic minorities, including immigrants, than other people, even after
controlling for political ideology and partisanship. Tea Party activists are also more likely to fear
falling in social class, more likely to be unemployed and more likely to view their taxes as being
unfair compared to both Tea Party supporters and the general public. Taken together these
findings suggest that those who participate in the Tea Party movement may be experiencing a
sense of threat to their cultural or economic statuses.
As has been discussed in the introduction and chapter 3, previous literature on right-wing
social movements has argued that right-wing movements emerge in response to threats perceived
by relatively (or formerly) high-status groups to their cultural, economic or political standing.
Much of this work has focused on local conditions associated with different forms of threat (e.g.
McVeigh 1999; Van Dyke and Soule 2002). Both McVeigh and Van Dyke and Soule show,
using state and county-level data in the United States, that conditions associated with particular
forms of threat at the local level were associated with local mobilization. Specifically McVeigh
(1999) found that local economic conditions and changing electoral characteristics were
associated with increased Klan membership, while Van Dyke and Soule (2002) show that
84
economic conditions, and the decline in manufacturing and family farming, were associated with
increased numbers of patriot militia organizations at the county and state-levels in the 1990s.
What much of this research leaves unaccounted for is how these local conditions interact
with larger, time-variant national political trends to produce local mobilization. After all, the
local conditions that produced a spike in Klan activity in the 1920s and a spike of patriot militia
activity in the 1990s were not unique to those time periods. The question then becomes, why do
the local conditions associated with economic threat sometimes lead to the mobilization of rightwing social movements and sometimes not. Traditional, structuralist theories of social movement
mobilization have argued that variation in mobilization can be accounted for by openings in the
political opportunity structure or the availability of resources. However, most scholars who study
movements on the right have rejected these explanations, noting that movements on the right
usually have high-status constituencies with ready access to resources and conventional political
channels. This leaves open the question of how local conditions interact with the broader
national political environment to produce the perception of threat and mobilization.
The remainder of this chapter will present evidence to support the theory that local
conditions interact with the national media and policy environment to produce local
mobilization. This chapter will argue that the existence of local conditions like increasing
unemployment or immigration are not necessarily enough, themselves, to generate right-wing
mobilization. Rather, the sense of threat must be activated by a national discourse that provides a
framework allowing individuals to perceive these changes as threatening.
The analysis in this chapter uses negative binomial regression to analysis to examine data
on Tea Party Patriot events collected by scraping the group’s event calendar between January
2009 and August 2011. Nearly 6000 events were successfully merged with a data set of state and
85
county-level economic and demographic variables. The findings indicate that the state and
county-level factors associated with Tea Party mobilization change over the course of the
movement, with immigration being important during the early stages of the movement when the
focus was on health care reform and economic issues becoming more important during the later
stages when the 2010 midterm elections became the focus of the movement.
WHO ARE THE TEA PARTIERS AND WHAT ARE THEY AFRAID OF?
In order to understand how the Tea Party movement might be responding to threat and
how this threat might interact with larger national environments over time, it is important, first,
to understand who Tea Partiers are – both their social position and cultural backgrounds. The
analysis from chapter 3 provides a good deal of insight on this issue, as does previous research
on the Tea Party movement. Tea Partiers tend to be older, white, men, more conservative and
more republican than average. Additionally, Tea Party members tend to view taxes as unfair and,
according to some previous research (e.g. Williamson et al.), view the unfairness of the tax
system primarily in a racialized context – that is they view their taxes as supporting undeserving
racial outgroups such as immigrants.
The existing literature on the Tea Party movement and the analysis in chapter 3 points to
two primary sources of threat that seem to motivate Tea Partiers on the individual level:
racial/cultural threat and economic threat. In this case racial threat seems to be particularly
focused on immigrants. Although, the analysis in chapter 3 found both anti-black and antiimmigrant sentiment to be higher among Tea Party supporters and Activists (even after
controlling for ideology and partisanship) than in the general public, much of the other research
on the Tea Party’s racial attitudes have found that immigrants, particularly illegal immigrants,
86
are the primary targets of Tea Party anger. Williamson et al (2011) found during qualitative
interviews with Tea Party activists, that activists often expressed animosity toward illegal
immigrants and believed that this group was unfairly benefiting from taxes, which they perceived
themselves as disproportionately paying. This finding is consistent with previous research on the
relationship between conservative attitudes toward race and ethnicity and opposition to welfare
state spending. For example, Gilens (1999) found that anti-black attitudes are a primary driver of
opposition to spending on social welfare programs in the United States, particularly when
welfare benefits are perceived as primarily benefitting blacks or when blacks are perceived to
disproportionately receive such benefits. Other research has found that the threat associated with
immigration also interacts with perceptions of the welfare state. Swank and Betz (2003) found
that immigration drives right-wing populist voting in European countries with non-social
democratic welfare state regimes. Additionally, Crepaz and Damron (2009) argue that welfare
states have the power to shape attitudes toward immigrants and that resentment of immigrants is
typically tied to a sense that immigrants benefit substantially from welfare state provisions
Additionally, Perrin, Tepper, Caren and Morris (2011) find that nationalism and
opposition to immigrants were one of the key cultural characteristics of Tea Party supporters in
the lead up to and after the 2010 Congressional elections. Futhermore, in their book, Parker and
Barreto (2013) argue that support for the Tea Party movement is rooted in cultural categories of
deservingness. According to the authors, Tea Party supporters view immigrants as a threat to
their understanding of what it means to be American.
Additionally, there is substantial evidence indicating that for many on the right, including
supporters and activists in the Tea Party, hostility toward immigrants is connected to attitudes
toward the welfare state, particularly health care reform. Many Tea Party supporters and activists
87
view health care reform as something that was likely to, in their view, unfairly benefit
immigrants. During the debate leading up to the passage of the ACA, many conservatives
expressed concern about the possibility that immigrants, particularly, illegal immigrants, would
gain access to benefits from the program. For example, Steve King, a Republican Representative
from Iowa stated in a press release in July, 20091:
Taxpaying families, already weighed down by bailouts and massive spending bills,
cannot afford to pay for health insurance for millions of illegal aliens. Hard and smart
working Iowans should not be forced to pay for illegal aliens to obtain health benefits
under any health care reform plan.
It is wrong to reward law breakers. The American people are speaking loud and clear and
saying, ‘No health care for illegal aliens.’
Anger on the right over this issue culminated most publicly during President Obama’s address to
a joint session of Congress regarding health care reform in September, 2009 when representative
Joe Wilson, a Republican from South Carolina shouted “you lie!” after the President stated that
illegal immigrants would not be covered under the plan. The public perception of a connection
between these two issues was so persistent, that three days later the White House was forced to
put out a statement clarifying, again, that Illegal Immigrants would not be covered under the
program.2 Even as recently as February, 2013 a poll found that nearly half of Americans
continued to believe that illegal immigrants would receive benefits under the health care reform
package.3 Figure 4.1 shows a plot of Google searches that contained both the words “health care”
and “illegal immigrants” in the same search4. Google search frequency is an effective way to
1
http://steveking.house.gov/media-center/press-releases/cbo-5600000-illegal-aliens-may-be-covered-underobamacare
2
http://abcnews.go.com/blogs/politics/2009/09/white-house-says-illegal-immigrants-will-be-explicitly-barredfrom-using-health-insurance-exchange/?cid=ESPNheadline
3
http://kff.org/disparities-policy/poll-finding/kaiser-health-tracking-poll-february-2013/
4
Similar analyses were conducted with other terms in place of “illegal immigrant” such as “illegals” and
“immigrants” and the substantive results were not different.
88
measure national discourse and public interest in issues as it is a measure of the extent to which
people actively seek out information on topics. Although, entering terms into a search engine is
not an expressive act, it does reveal the extent to which the topics being searched are a concern
to members of the public and is very likely to reflect issues that are central to the national
discourse (Stephens-Davidowitz 2014; Swearington and Ripberger 2014). As is clearly evident
in the plot, joint interest in these two topics was reduced to close to 0 shortly after the passage of
the ACA, providing strong evidence that illegal immigration was closely tied in the minds of
many with health care reform.
[FIGURE 4.1 ABOUT HERE]
Previous research also indicates that Tea Partiers are concerned about threats to their economic
status. The analysis in chapter 3 shows that Tea Party members are more likely to be
unemployed, more likely to be concerned about downward mobility and more likely to consider
their taxes unfair. Additionally, some journalistic evidence supports these findings. Zernicke
(2010), in her study of the early Tea Party movement, found that many people brought up recent
job loss, unemployment or other financial problems when discussing their reasons for deciding to
participate in the movement. Of course, it is also likely that Tea Party members do not see
racial/cultural threat as being entirely separate from economic threat, as previous research on
conservative movements has shown.
Why would we expect the nature of threats to change over time?
There is an existing literature in the field of social movements that emphasizes the simple
idea that movements adapt themselves to political events and changing social contexts with a
number of scholars arguing that movements alter their tactics, targets, mobilization structures,
89
strategies and alliances over time as the social context in which they operate changes (Tarrow
1993; Staggenborg 1998; Tarrow 1998; Blee and Currier 2006). This literature suggests that
changes in the political opportunity structure are responsible, not only for attracting activists and
increasing mobilization, but also for changing the goals and nature of a movement over the
course of a protest cycle (Staggenborg 1998; Tarrow 1998). As mobilization increases and
opportunities shift, movements develop new strategies, modes of contention, frames and targets
to adjust to their changing environment; as such, movements are not static.
Indeed, Fligstein and McAdam (2011) argue that the boundaries of a movement, itself,
can change in response to changing external opportunities and threats. Blee and Currier (2006)
find that social movement groups react to changes in their political environment, particularly
elections, by altering their strategies and goals. Social movement groups often see elections as
opportunities to advance their goals, attract new members or forge new alliances. While this
literature is insightful, it does not account for the unique characteristics of right-wing
mobilization; particularly how the perception of threat interacts with a changing national
environment. Although other research has found that the perception of threat can change in
response to changes in the national context, this work has not looked at this process from the
context of the localized perception of threat which has been found to be key to right-wing
mobilization (Heaney and Rojas 2011). Heaney and Rojas (2011) found that the anti-Iraq War
movement demobilized after the election of Barack Obama made war policies seem less
threatening to Partisan Democrats in the movement.
The experience of threat requires not just the existence of certain structural conditions or
changes, but also the perception of threat. The perception of threat (as experienced locally
through local conditions and events) is conditioned by the broader, national political
90
environment. Therefore, it is not just that an increase in immigration causes threat, the salience
of immigration has to be increased by the national context. Previous research on anti-immigrant
policies has produced findings consistent with this idea. Hopkins (2010) finds that perception of
immigrants as threatening is produced through the interaction of national and local conditions.
He argues that the key to the perception of threat from immigrants (and resultant passage of local
anti-immigrant policies) occurs when there is a local influx of immigrants at the same time that
national rhetoric on immigration (in the form of the national media) provides a salient frame
casting immigration as threatening.
I will argue that a very similar process produces the perception of threat responsible for
right-wing Mobilization in the Tea Party movement. For example, the threat of status decline
associated with race, immigration, job loss or taxes may motivate Tea Party mobilization,
however, experience of local conditions associated with such threats, itself, does not cause rightwing mobilization. Immigration, taxes, and job loss existed before February 2009 and did not
lead to mobilization. Instead it requires the interaction between the national political and media
environment and the local conditions associated with threat (e.g. increasing immigrant
populations, high unemployment) to produce the perception of threat and mobilization on the
local level. In other words, national rhetoric can politicize people’s day to day environments
when these environments might otherwise have gone unnoticed or have not been interpreted in a
politicized manner.
The Changing Context of Tea Party Activism
The Tea Party movement was highly active from early 2009 through the 2010 midterm
elections into 2011. The Tea Party Patriots, a network of approximately 3000 local Tea Party
91
chapters, was particularly active in organizing meetings, local protests and other events 5. During
this time period the threats and opportunities encountered by the movement as it navigated the
political environment of the U.S. changed considerably. During the early period of the
movement, debate over the Affordable Care Act (ACA) dominated the political discourse. The
Tea Party led opposition to this legislation, perhaps most visibly with its “town hall” protests
during the summer of 2009. Tea Party activity around the ACA continued until the passage of
the bill in March 2010. Shortly afterwards the national discourse shifted focus to the 2010
midterm election season. During this time period the Tea Party became an influential force in
electoral politics and shifted the focus of its activities toward support for conservative
Republican candidates. To illustrate this change in the national discourse, Figure 4.2 shows
normalized Google search frequencies over time for the terms “health care” and “election. 6”
[FIGURE 4.2 ABOUT HERE]
Figure 4.2 clearly shows searches for “health care” as being predominant in the early period of
the plot and subsiding after the passage of the ACA in March of 2010, shortly after which
searches for the term “election” increased dramatically.
The changing social context the Tea Party operated in over this time period provides an
ideal opportunity to analyze how Tea Party mobilization changed as the social and political
context surrounding the movement changed. The analysis will evaluate how the movement’s
mobilization base, as well as the forms of threat that are associated with mobilization changed in
response to this changing context. During the health care period different forms of threat are
5
This figure is based on data compiled by scraping the Tea Party Patriots website
(http://www.teapartypatriots.org/local/) on September 16, 2011. Note: the architecture of the website has
changed substantially from when it was originally scraped in September 2011.
6
Data on Google search frequencies obtained from google.com/insights/search
92
likely to have been more salient and different constituents are likely to have been mobilized than
during the election period.
For example, consistent with previous research on racialized opposition to the welfare
state, we may expect racial fears and threats among whites to have been more salient during the
health care reform period – a period during which the nation was considering an expansion of the
welfare state. This is particularly likely given the apparent connection between the antiimmigrant sentiment and health care reform in Tea Party rhetoric. Thus, structural conditions
associated with racial threat may have been more influential during this time period than during
the election. Conversely, during the 2010 election period, the focus on issues related to national
austerity during the 2010 election campaigns, highlighted in Figure 4.3, may have made
economic concerns among conservatives more salient. In this sense, we may expect wealthier
counties – those that may be more threatened by the prospect of taxation – to have higher rates of
Tea Party activity during the election season.
[FIGURE 4.3 ABOUT HERE]
Figure 4.3 clearly shows that terms related to austerity, tax cuts and budget cuts exploded during
the election period as compared to the health care period. Given that the campaign rhetoric
focused on these issues, high income individuals who felt their economic status was threatened
may have found this rhetoric particularly salient. It should be noted, that I am not arguing that
health care and election issues were the only issues the Tea Party focused on nor that health care
and austerity related issues were the only issues to emerge in the national political discourse
during this time. However, these issues were the dominant political debates over the time frame
of this study and provide good cases to test the effect of discourse on the perception of local
threat.
93
Hypotheses and Expectations
Three general hypotheses are derived from the previous literature:
H1: The effect of local sources of threat will not be consistent over the course of the movement
and will fluctuate as changes in the national media and policy environments politicize different
forms of local threat.
H2: Consistent with previous research, counties and states with conditions that are associated
with out-group threats, specifically international immigration and percentage of the population
that is black are expected to have higher Tea Party event counts. This effect is likely to be
particularly strong during the Health Care reform period because group threat is frequently
associated with racialized views of the welfare state.
H3: Consistent with previous literature there is expected to be an effect associated with measures
of economic threat such as median household income and change in unemployment rate. Higher
income households and those who feel they are at risk of downward economic mobility might be
particularly hostile to the prospect of taxation government spending. Given the increased focus
on these issues during the 2010 election period, it is likely that the effect associated with
economic threat measures will be greater during this period.
DATA AND METHODS
94
The following analysis will test the above hypotheses using a county-level and state-level data
set with a dependent variable constructed from a record of all official Tea Party patriot events in
the United States from between January 2009 and August 20117. The data used to construct the
dependent variables were obtained by scraping the Tea Party Patriots website with an automated
script that recorded the time of each event as well as the self-reported location of the event8. The
events in the data were then matched to counties and states by a script, which was able to match
the majority of events to U.S. counties or county equivalents. Some of the locations listed on the
Tea Party event calendar were either too general (e.g. “All of Nebraska” or “Your Town”) or
were inscrutable and were thus unable to be matched to a county. In the end 5,851 of the 6,193
events were successfully matched to a U.S. county including 2,480 during the health care time
period and 3,371 during the election period.
Dependent Variable
These events recorded in the construction of the dependent variable include a broad array
of activities such as protests as well as more mundane events such as meetings, talks and
speeches. The dependent variable, shown plotted over time in Figure 4.4 is consistent with the
history of Tea Party activism. A clear peak is visible in late summer and early fall of 2009,
representing the period in which Tea Party “town hall” protests against health care reform where
taking place. We see activity escalate again the months leading up the passage of the ACA in
7
Although, the data collection period technically begins in January 2009, very few events are recorded before
August of 2009.
8
Web scraping is a data collection technique in which a computer program is used to automatically obtain machine
readable data from text on internet web pages. Generally, a script is written to automatically load webpages and
then extract and parse the text. Relevant portions of the text can then be automatically coded by the script into
datasets amenable to statistical analysis. The data is obtained from an older version of the Tea Party Patriots event
calendar: http://www.teapartypatriots.org/events/. The script used to obtain the data is written in Python and is
available in the appendix at the end of chapter 2.
95
March 2010, followed by a large spike in April of 2010 in which activists participated in the Tea
Party’s April 15th tax day protests. Shortly after the passage of the ACA, Tea Party activity
declined and then proceeded to increase again in the months leading up to the 2010
Congressional election.
[FIGURE 4.4 ABOUT HERE]
In the county-level analysis, three different operationalizations of this measure are
employed in the analysis as dependent variables: a measure of Tea Party Patriot events during
the time period in which the national discourse was focused on health care reform, a measure of
Tea Party Patriot events from the period in which the national discourse was focused on the 2010
midterm elections and a pooled variable including events from both time periods. The health care
reform period extends from January 2009 through May 2010 while the election time period
covers June 2010 through the post-election period to August 2011. This division is consistent
with a natural breaking point in the history of the movement – it includes the run up to the
passage of the ACA as well as the post ACA “tax day” protests in the health care period while
including all the activity after that point in the election period. The division is also consistent
with the prevalence of national discourse on health care reform and the election shown in Figure
4.2. It is clearly evident in Figure 4.2 that the popularity of the search term ‘health care’ dies
down near the division point and the popularity of the term ‘election’ increases rapidly after that
point, having several distinct peaks in the following months. In the state-level analysis, the
events are matched to states in state-month format to allow for time-series cross-sectional
analysis.
96
State and County-Level Independent Variables
The data used in this analysis consist of county and state-level dataset compiled from a number
of sources including the County Characteristics, 2000-2007 dataset compiled by the InterUniversity Consortium for Social and Political Research as well as more recent data on county
unemployment rates from the Bureau of Labor Statistics and state-level data from the U.S.
census bureau9. The county-level data are summarized below in table 4.1.
[TABLE 4.1 ABOUT HERE]
County-Level Independent Variables
Independent variables in the county-level analysis include measures of unemployment
change between 2007 and 2009, median house hold income (2010), international immigration
into the county, percent of the population that is black and John McCain’s vote share from the
2008 presidential election. Additionally, control variables for county population, dominant
economic sector, median age and whether or not the county is urban or rural are included.
Unemployment Change
A measure of the change in the unemployment rate between 2007 and 2009 is included as an
indicator of the extent to which a county was impacted economically by the financial crisis. The
measure was constructed by subtracting the unemployment rate in 2007 from the unemployment
rate in 2009, yielding positive values when unemployment increased between 2007 and 2009 and
negative values in the instances where unemployment decreased between 2007 and 2009. The
data used to construct this measure comes from the Bureau of Labor Statistics.
9
https://www.census.gov; http://www.bls.gov/lau/
97
Median Household Income
A measure of 2010 median household income divided by 1,000 is included as a measure of the
affluence of each county. This measure is intended to be an indication of the extent to which the
population of a county may perceive an economic threat from taxation and government spending.
Dominant Economic Sector
Another set of variables relevant to the economy, included primarily as a control, are measures
of the dominant economic sector in each county. Included are dummy variables for
manufacturing, mining, farming, government and service sector. Non-specialized counties serve
as the reference category. These items come from the County Characteristics, 2000-2007 data
set.
International Immigration
In order to assess the effect of out-group threat, a measure of the foreign immigration rate in each
county is included. The included measure is the total net international migration between July
2004 and July 2005 as a proportion of county population. This measure gives an indication of the
extent of immigration into the county in the time period immediately preceding the financial
crisis and the emergence of the Tea Party movement.
Percent Black
Another measure of out-group threat is percentage of the population that is black. Included in the
model is a linear and quadratic term for percent black. This specification is intended to pick up
98
non-linear relationships between percent black and Tea Party mobilization. Previous research has
indicated that political activity motivated by racial threat often follows a non-linear relationship
(Blalock 1967; Taylor 1998).
Percent McCain Vote Share
A measure of the percent vote share received by John McCain in the 2008 Presidential election is
also included in the analysis as an indicator of the partisan breakdown of the county. This
measure will allow for an examination of the role of partisan politics in Tea Party mobilization.
Also included in the analysis is a quadratic term for percent McCain vote in order to account for
potential non-linearities in the relationship between McCain vote share and Tea Party
mobilization.
State-Level Independent Variables
[TABLE 4.2 ABOUT HERE]
State-Level independent variables are shown below in Table 4.2. The dependent variable at the
state-level is a binary variable, with a value of one indicating that the state did have a Tea Party
event that month and a zero indicating that it did not. The measure for state population is based
on the US census 2009 population estimate for each state. The %McCain vote measure is
intended to measure partisanship and serve as a baseline measure of the conservatism of each
state. It is constructed from data provided by the Federal Election Commission (FEC). The
Unemployment Change measure, like its county-level counterpart is intended measure the impact
of the financial crisis on the economic conditions in each state. It is constructed by subtracting
the 2007 unemployment rate from the 2009 unemployment rate. This data comes from the
99
Bureau of Labor Statistics (BLS). The immigration rate measure consists of the number of
immigrants to each state between 2000 and 2009 (in thousands). This data comes from estimates
produced by the U.S. Census Bureau. The %Urban measure represents the percentage of each
state’s population living in Urban areas; this data comes from the U.S. Census Bureau. The
median wage item, intended to measure the income of the residents of each state, comes from the
Bureau of Labor Statistics. Finally, the time variable is measured in months over the period from
January 2009 to August 2011. This period covers 32 months.
In addition to the state-level variables, eight national discourse measures based on
Google search frequencies and Fox News coverage are employed in the analysis. The Google
search terms include nationwide measures for the terms ‘Health Care,’ ‘Health Care’&’Illegals,’
‘Big Government,’ and ‘Tax Cuts.’ Additionally four items based on Fox News coverage
employed in the analysis are based on the terms ‘health care,’ ‘Obamacare,’ ‘Deficit,’ and
‘Austerity.’ These items consist of counts of the number of articles on foxnews.com that
mentioned each term in a particular month10. The discourse variables are time variant, but do not
vary between states.
Analytical Strategy
The analysis in this chapter will make use of both county and state-level data to test the
hypothesis suggested by the previous literature. First, the data will be modeled at the countylevel using negative binomial regression analysis. County-level analysis allows for a
geographically fine grained analysis, but is not amenable to time-series modeling due to
convergence problems and data sparsity. For this reason, the county-level analysis will be limited
10
This data was collected by searching the Google News aggregator for Fox News articles containing these terms in
the designated time period.
100
to two time periods – the health care period and election period. Three separate county-level
models are run. First a pooled model is run with both time periods, allowing a time dummy
variable to be interacted with independent variables of interest to test for changes in the effect of
the independent variables across time periods. Second, separate models will be run for each time
period to confirm the results from the pooled analysis. The state-level analysis will be run as a
time-series cross sectional analysis, in which variables measuring national media discourse will
be interacted with state-level independent variables to measure the extent to which state-level
independent variables are affected by changes in the national media discourse. Although the
state-level analysis does not allow for the geographical granularity of the county-level analysis, it
does allow for the effect of national discourse on independent variables to be tested directly.
County-Level Modeling Strategy
Given that the dependent variable used in the county-level analysis is a count outcome,
conventional ordinary least squares regression models can produced biased, inconsistent and
inefficient estimates (Long 1997). Count outcomes are frequently modeled using Poisson
regression models, or in the presence of overdispersion, negative binomial regression. As a
likelihood-ratio test provides strong evidence of overdispersion, this analysis employs negative
binomial regression. To model the pooled dependent variable, negative binomial models with
county-level clustered standard errors are employed in order to account for within-county
correlation of errors between the measurements at the two time points included in the data.
Interaction effects between the time variable and several of the independent variables are
included in the model to statistically test whether the effect of the variables change from time
one to time two. Such interactions are included for unemployment change, median household
101
income, and international migration. As the time variable is a dummy variable coded as 1 for the
election time period and 0 for the health care time period, interaction effects can be interpreted
such that the main effect is the effect for the variable during the health care period and the main
effect plus the coefficient for the interaction term is the effect for election time period. In the
interest of robustness and clarity models for the two time periods are also estimated separately
using negative binomial regression.
State-Level Modeling Strategy
The state-level analysis is a cross sectional time series (CSTS) analysis using a binary dependent
variable with time points nested in states. In order to test the effect of national discourse on local
determinants of mobilization, local factors will be interacted with media measures. The models
are defined by the following equation
𝑘
𝑦𝑖𝑡∗
= 𝛼 + ∑(𝛽𝑘 𝑥𝑘𝑖 ) + Γ𝑧𝑡 + Ρ𝑥𝑖 𝑧𝑡 + 𝜁𝑖 + 𝜖𝑖𝑡
1
Where 𝑧𝑡 represents a time-variant national discourse variable, 𝑥𝑖 -s represent state-level
independent variables and 𝑥𝑖 𝑧𝑡 represents the interaction effect.
Ordinarily, such models are prone to a number of endogeneity problems, including the
possibility that the error component is correlated with an explanatory variable. However, because
in the models estimated here, the only time variant variables are the dependent variable and
national media variables, which do not vary by state, this is not an issue.
RESULTS
102
The results of the county-level negative binomial regression analyses are shown below in Table
4.3. Column one represents the results of the pooled model, while columns two and three show
the results of the health care model and the 2010 election model, respectively.
[TABLE 4.3 ABOUT HERE]
On the whole, the three hypotheses are largely confirmed by the data in the county-level
analysis. Out-group threat is found to be associated with Tea Party Mobilization and the effect of
international immigration is largely limited to the health care reform period. An effect for
economic threat is also observed in the data, with the change in unemployment rate after the
financial crisis having an effect in both time periods and higher income having an effect only
during the election period. Additionally the partisan competitiveness of the county is shown to
have an effect, however, this effect is not significantly stronger during the election time period
than during the health care reform period.
With respect to the economic measures included in the analysis, it is evident that
increased unemployment between 2007 and 2009 is associated with higher levels of Tea Party
mobilization during both the health care and election time periods. This finding seems to indicate
that Tea Party mobilization does respond to economic threat – counties that were hit harder by
the financial crisis and recession were more likely to have Tea Party events holding all other
factors constant. The effect during the election time period is substantially larger than the effect
during the health care time period, as is evident in the positive interaction term in the pooled
model and the larger coefficient in the individual model for the election period. While this
difference falls short of statistical significance at conventional levels, it does suggest that the
effect of unemployment may have been greater during the election period than during the earlier
health care period. Although the effect for unemployment rate is significant, it is relatively
103
modest. According to the pooled model, during the health care period, an increase in
unemployment rate change from one standard deviation below the mean to one standard
deviation above is associated with an increase in the probability of a non-zero event count from
.23 to .28, holding other variables at their mean.
Furthermore, there is a strong positive effect for median household income during the
election period, but there appears to be no significant effect during the health care period. This
indicates that higher income counties were more likely to be the location of Tea Party events
during the 2010 election cycle but were not more likely to be the location for such events during
the lead up to health care reform. The effect of median household income for each time period is
shown below in Figure 4.5.
[FIGURE 4.5 ABOUT HERE]
Holding all else equal, a change from one standard deviation below the mean to one standard
deviation above the mean for median household income during the election period accounts for
an increase in the probability of a non-zero Tea Party event count of approximately .034.
However, the probability of a non-zero event count is nearly .16 higher for the wealthiest
counties than it is for the poorest counties. This may be indicative of wealthier counties
experiencing more economic threat during the election period when issues of taxes and
government spending were being debated.
Additionally, it should be noted that according to the pooled model, counties whose
economies were dominated by farming, manufacturing and mining based economic sectors
associated with producing large numbers of working class jobs, were less likely to have non-zero
Tea Party event counts than counties dominated by the service sector. It should also be noted that
the effect for mining is only marginally significant, narrowly falling short of significance at the
104
.05 level in the pooled model, though it does have a negative, significant effect in the separate
health care and election models.
With respect to status group threat, there is a positive effect for immigration during the
health care reform era, however this effect is greatly reduced and becomes insignificant during
the 2010 election period. This is evident from the large and significant negative interaction term
in the pooled model as well as the lack of a significant coefficient in the separate 2010 election
model. This finding indicates that higher levels of immigration were positively associated with
Tea Party mobilization during the health care reform debate, but not during the 2010 election
cycle. In other words, international immigration seems to have been a mobilizing factor during
the health care reform debate but became unimportant as a predictor of Tea Party event counts
during the election time period. The effect of immigration on Tea Party mobilization is shown
below in Figure 4.6.
[FIGURE 4.6 ABOUT HERE]
During the health care period, there is an increase in the predicted probability for a non-zero
event count of .062 associated with a change in the immigration rate of one standard deviation
below the mean to one standard deviation above, holding all else constant. Counties with the
highest immigration rates have a probability of a non-zero event count that is nearly .27 higher
than those with the lowest immigration rates.
The percentage of the population that is black is also a potential measure of cultural or
out-group threat. There is a curvilinear effect for the variable percent black, with the highest
level of Tea Party mobilization occurring at approximately 20% percent black. After peaking at
around 20%, the probability of a non-zero event count drops rapidly to near 0 for the counties
that have the highest percentage of black residents. No significant differences in the effect of this
105
variable were found between the two time periods.11 This finding is consistent with previous
literature on racial threat theory, which has found that an inverted-U shaped curvilinear
relationship between a dependent variable and percent minority is an indication of a sense of
threat from the minority group (Blalock 1967; Talyor 1998). Specifically, Taylor (1998) found
an inverted U-shaped relationship between prejudiced attitudes and the size of the minority
population as a percentage of the total population of an area.
[FIGURE 4.7 ABOUT HERE]
The effect of partisanship, measured as McCain vote share in 2008, is strongly
curvilinear. This relationship is show below in Figure 4.8.
[FIGURE 4.8 ABOUT HERE]
Counties with approximately 50% McCain vote shares have the highest probability of non-zero
Tea Party event counts. There is no statistically significant difference in this effect between the
health care reform period and the election period. 12 Tea Party events appear to be most likely to
occur not in the most conservative counties, but rather those that are the most competitive –
those in which the sense of political threat for conservatives is highest and those in which the
potential for political victory is most uncertain. Interestingly, no significant difference was found
in the effect of percent McCain between the time periods.
The results of the state-level time series cross sectional binary logit analysis using health
care related discourse terms are shown below in Table 4.4.
[TABLE 4.4 ABOUT HERE]
11
Interaction terms between percent black and percent black squared and the time dummy variable, not shown in
the table, did not show significant differences between the two time periods.
12
Interaction terms between percent McCain vote and percent McCain vote squared with the time dummy
variable were not significant. Interaction terms not included in model shown.
106
Across all models, median wage has a positive and significant effect on Tea Party mobilization
while the main effect of immigration has a negative, significant effect. However, as indicated by
the interaction effects, the total effect of immigration is not negative for all values of the national
discourse variables. In model one, Google search frequency for the term ‘health care’ is
interacted with immigration. Here the effect of the interaction is positive, indicating that at
higher values of the health care discourse measure, the effect of state-level immigration becomes
larger. Thus, at higher values of health care discourse, the net effect of immigration becomes
positive. Additionally, the main effect of health care discourse is positive. An alternative
interpretation of the interaction effect is that the effect of health care discourse is larger in states
with larger numbers of recent immigrants net of other factors.
Column two, in table 4, shows results that are similar to those in column one. In column
two, the joint Google search frequency for “healthcare” and “illegals” is interacted with
immigration. Although the main effect for immigration is negative, the interaction term is
positive and the net effect of immigration becomes positive at higher values for the search term.
Additionally, the main effect of the search term is positive.
Columns three and four show models using discourse measures constructed from Fox
News articles. Columns three interacts the Fox News ‘health care’ measure with immigration and
the results are consistent with the Google search results. Although the main effect of immigration
is negative the interaction effect is positive indicating that at higher values for the Fox News
measure the net effect of immigration is positive. Additionally, the main effect of the Fox News
‘health care’ measure is positive and significant. Column 4 shows the results of a model
interacting a Fox News metric based on the term ‘Obamacare’ with the immigration variable.
Again, while the main effect of immigration is negative, the interaction effect is positive,
107
indicating that at higher levels of the ‘Obamacare’ variable, the net effect of immigration
becomes positive. Additionally, the main effect of the ‘Obamacare’ variable is positive and
significant. Taken together, these results clearly show that, net of other factors, Immigration as a
larger, positive effect on Tea Party mobilization at time periods when the national discourse was
more heavily focused on health care reform.
Table 4.5 shows the results of models interacting national discourse variables relating to
the size of government and taxation with the median wage variable.
[TABLE 4.5 ABOUT HERE]
The model in column one interacts a Google search frequency measure for the term ‘Big
Government’ with the median wage. The main effect for median wage is insignificant, however
the interaction effect is positive and significant, indicating that time periods during which the
national discourse was more heavily focused on ‘Big Government,’ the effect of the median
wage on Tea Party mobilization was greater13. The main effect for ‘Big Government is negative
and significant. A model for using a Google search item for the term ‘tax cuts’ is shown in
column 2. Here the main effect of median wage is positive and significant, indicating that higher
income states were more likely to have Tea Party events net of other factors. However, the
interaction effect with the Google search term is not significant.
Columns three and four show the results of the Fox News items for the terms ‘deficit’ and
‘austerity,’ respectively. In column three, the main effect for median wage is insignificant,
however the interaction effect is positive and significant, indicating that at times when more Fox
News coverage contained the term ‘deficit,’ the effect of median wage on mobilization was
greater. The mean effect for the ‘deficit’ item is negative and significant. In column 4, a Fox
13
Note: ‘Big Government (biggovernment.com)’ is also the name of a website popular with conservatives. It is
likely that some of the search traffic related to this term is referring to the website.
108
News item for the term ‘austerity’ is interacted with median wage. In model four neither the
main effect of median wage, nor the interaction effect with the ‘austerity’ term are significant.
Taken together these models provide some evidence that discourse surrounding issues related to
the size of government increases the effect of median wage on Tea Party mobilization, however,
the evidence is mixed and not as strong as the evidence for the relationship between health care
discourse and immigration.
CONCLUSIONS
The results of the analyses in this chapter provide evidence that local factors relating to
economic and racial/cultural threat are important predictors of Tea Party mobilization. Economic
factors include unemployment and household income and cultural factors include immigration
and racial composition. Additionally, other factors were also found to be significantly associated
with Tea Party mobilization such as the partisan composition of counties, with more evenly
divided counties having more Tea Party events. However, the importance of different forms of
threat varied over the course of the movement. The county-level analysis shows that immigration
is an important predictor of Tea Party mobilization during the early part of the movement, when
the debate over health care reform dominated the national discourse, but was not a significant
predictor during the later period. This finding, which suggests that the healthcare debate made
racialized opposition to an expansion of the welfare state more salient, is further supported in the
state-level analysis which shows that the effect of immigration at the state level is greater during
time periods when the national discourse was more heavily focused on health care reform.
Additionally, the county-level analysis shows that economic threat measures are
important during both time periods, but seem to be more important during the 2010 election
109
period than during the earlier health care reform period. This was particularly true with respect to
household income. This suggests that a national discourse focused on election issues, such as the
size of government, were making local sources of economic threat more salient. Higher income
individuals may have perceived taxation and government spending to be a greater threat to their
economic status when the national discourse focused on these issues. However, the state-level
analysis provided only mixed support for this hypothesis. Thus, while economic factors clearly
did become more important during the election, the reason for this change remains somewhat
more ambiguous.
There are a number of reasons why economic factors may have been more important
during the later stages of the movement. For example, all else being equal, residents of poorer
counties likely stood to benefit more from an expansion of health insurance than those in
wealthier counties. This may have served to blunt opposition in poorer areas during the first time
period. Additionally, as demonstrated in chapter 3, fewer Americans supported the Tea Party
during the later stages of the movement and those who remained were more partisan and more
conservative. It is possible that the demographics of the movement shifted in ways that made
more traditionally Republican groups, such as those with higher incomes, more central to the
movement.
Taken together the results of this analysis show the importance of a changing political
environment and social context on Tea Party mobilization. The Tea Party is a coalition of
different constituencies who share resources and an identity, but respond to different political
events and are potentially activated by different forms of threat. When the nation was debating
an expansion of the welfare state through the Affordable Care Act, racial/cultural threat was the
most important mobilizing factor and Tea Party mobilization was concentrated in areas with high
110
immigration rates where this form of threat would be most salient. During the election period,
when the national discourse was dominated by talk of austerity and taxation, economic threat
seemed to be particularly important. Tea Party events were especially concentrated in high
income areas, where the residents would be most sensitive to economic threats from government
spending and taxation. Mobilization, therefore does not necessarily arise from threatening social
conditions alone, rather these conditions must be politicized and made salient by a broader
national political environment.
111
Table 4.1: Descriptive Statistics for county-level data (N=3112 Counties)
Standard
Variable
Mean/Proportion Dev.
Minimum Maximum
Dependent Variables
Events (Health Care Period)
0.79
2.4
0
45
Events (Election Period)
1.07
3.56
0
75
Economic threat
Unemployment Change
4.15
2.11
-1.6
15.3
Median Household
Inc./1000
43.13
10.74
20.58
119.08
Out-Group Threat
Intl. Migration 05
0.12
0.18
0
1.45
% Black
8.99
14.54
0
85.99
% Black Squared
292.24
791.12
0
7394.28
Political Competitiveness
% McCain
56.89
13.84
6.54
93.21
% McCain Squared
3428.17
1532.58
42.77
8688.104
Control Variables
Manufacturing
0.29
N/A
0
1
Mining
0.04
N/A
0
1
Farming
0.14
N/A
0
1
Government
0.12
N/A
0
1
Service Sector
0.11
N/A
0
1
Urban
0.35
N/A
0
1
Rural
0.39
N/A
0
1
Median Age
38.61
4.43
20.1
55.3
Population/1000
94.37
306.43
.06
9935.48
Time
.5
N/A
0
1
Notes:
Data compiled from the County Characteristics Survey 2000-2007, the Bureau of Labor Statistics
and the Tea Party Patriots online event calendar.
112
Table 4.2: Descriptive Statistics for state-level variables.
Variable
Events
Mean
StdDev
Minimum
Maximum
0.56
N/A
0
1
6128.14
6738.42
544.27
36961.66
%McCain
Unemployment
Change
47.84
9.4
26.4
65.6
4.1
1.42
1.2
7
Immigration Rate
178.4
313.4
3.04
1816.63
%Urban
71.69
14.76
38.18
94.44
Median Wage
42.08
6.44
32.82
60.29
Time (month)
16.5
9.24
1
32
Population
113
Table 4.3: Negative binomial regression results for effect of independent variables on Tea Party Events.
Pooled
Health Care
2010 Election
Economic Threat
Unemployment Change
0.080**
0.075**
0.139***
(2.85)
(3.04)
(4.72)
Time × Unemp Change
0.051
(1.43)
Median HH inc./1000
-0.001
-0.002
0.013*
(-0.30)
(-0.38)
(2.20)
Time × Median HH Inc.
0.013*
(2.31)
Out-Group Threat
Immigration 05
1.285***
0.972**
0.512
(3.85)
(2.91)
(1.32)
Time × Immigration
-1.009**
(-3.29)
Percent Black
-0.241**
-0.191*
-0.285**
(-2.82)
(-2.24)
(-2.70)
Percent Black Squared
-0.271*
-0.329**
-0.219
(-2.36)
(-2.92)
(-1.54)
Political Competitiveness
Percent McCain
-0.008*
-0.009*
-0.007
(-2.02)
(-2.31)
(-1.64)
Percent McCain Squared
-0.0006**
-0.0003
-0.0008**
(-2.86)
(-1.78)
(-3.26)
Control Variables
Manufacturing
-0.275**
-0.199
-0.381**
(-2.58)
(-1.83)
(-2.98)
Mining
-0.556
-0.533*
-0.562*
(-1.92)
(-1.99)
(-2.09)
Farming
-1.542***
-1.398***
-1.697***
(-5.65)
(-5.92)
(-6.92)
Government
0.245
0.233
0.237
(1.77)
(1.72)
(1.48)
Service Sector
0.268*
0.422**
0.105
(2.35)
(3.19)
(0.64)
Urban
0.464***
0.618***
0.344**
(3.99)
(5.49)
(2.60)
Rural
-0.625***
-0.537***
-0.697***
(-4.53)
(-4.41)
(-5.06)
Median Age
-0.004
-0.015
0.006
(-0.34)
(-1.18)
(0.40)
Time
-0.345
(-1.01)
Population/1000
0.002***
0.001***
0.002***
(4.80)
(5.89)
(5.68)
Constant
-0.929
-0.510
-1.717*
(-1.65)
(-0.92)
(-2.48)
Ln(alpha)
1.264***
0.935***
1.497***
(22.06)
(14.13)
(26.74)
N
6224
3112
3112
Notes:
z statistics in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
114
Table 4.4: Panel Logistic Regression Results for Media Terms Interacted with Immigration (N=1600)
Variable
State Population
G: Health Care
0.0004***
(5.643)
McCain Vote 08
Unemployment Change
Immigration Rate
%Urban
Median Wage
Time (month)
G: HC + Illegals
0.0004***
(5.870)
FN Health Care
0.0004***
(5.795)
FN: Obamacare
0.0003***
(5.575)
0.034
0.034
0.033
0.029
(1.868)
(1.837)
(1.852)
(1.875)
0.094
0.084
0.08
0.048
(0.801)
(0.697)
(0.702)
(0.494)
-0.010***
-0.008***
-0.010***
-0.006***
(-5.410)
(-4.826)
(-5.303)
(-4.693)
0.003
0.004
0.003
-0.001
(0.213)
(0.240)
(0.239)
(-0.118)
0.072*
0.073*
0.069*
0.058*
(2.061)
(2.030)
(2.030)
(1.974)
0.162***
(16.750)
0.174***
(17.170)
0.148***
(15.862)
0.101***
(11.236)
Google Search
Immigration × Health Care
0.014**
(2.831)
Health Care
0.146***
(6.877)
Immigration × Illegals+HC
0.008**
(2.674)
Illegals + Health Care
0.110***
(9.914)
Fox News
Immigration × Health Care
0.007*
(2.565)
Health Care
0.035***
(8.473)
Immigration x Obamacare
0.020***
(4.303)
Obamacare
-0.006*
(-1.990)
Constant
-10.601***
-10.087***
-11.118***
-6.427***
(-5.387)
(-5.065)
(-5.731)
(-3.974)
-0.276
-0.219
-0.335
-0.722*
(-1.135)
(-2.277)
lnsig2u
_cons
(-0.936)
(-0.752)
Notes: z statistics in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001
115
Table 4.5: Panel Logistic Regression Results for Media Terms Interacted with Immigration (N=1600)
Variables
State Population
McCain Vote 08
Unemployment Change
Immigration Rate
%Urban
Median Wage
Time (month)
G: Big Gov
0.0004***
(5.027)
-0.045**
(-2.926)
0.016
(0.155)
-0.006***
(-3.729)
0.004
(0.321)
-0.023
(-0.642)
0.104***
(12.221)
G: Tax Cuts
FN: Deficit
FN: Austerity
0.0003***
(4.768)
-0.043**
(-2.900)
0.037
(0.361)
-0.005***
(-3.722)
0.006
(0.460)
0.073*
(2.427)
0.138***
0.0003***
(4.865)
-0.041**
(-2.911)
0.024
(0.247)
-0.005***
(-3.723)
0.005
(0.407)
-0.007
(-0.204)
0.107***
0.0003***
(4.805)
-0.042**
(-2.900)
0.032
(0.323)
-0.005***
(-3.727)
0.006
(0.444)
0.047
(1.624)
0.146***
(15.380)
(11.343)
(14.778)
Google Search
Wage x Big Government
Big Government
0.536***
(4.735)
-15.204***
(-3.463)
-0.065
(-1.455)
0.811
(0.420)
Wage x Tax Cuts
Tax Cuts
Fox News
0.166**
(2.785)
-6.008*
(-2.432)
Wage x Deficit
Deficit
-2.572*
(-1.975)
-5.528***
(-5.133)
-2.343
(-1.730)
0.117
(1.760)
-7.083**
(-2.582)
-4.665***
(-4.560)
-0.489
(-1.582)
-0.569
(-1.844)
-0.700*
(-2.228)
-0.611*
(-1.969)
Wage x Austerity
Austerity
Constant
lnsig2u
_cons
Notes: z statistics in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001
116
Figure 4.1: Google search frequency for joint searches including “illegal immigrants” and
“health care.”
5
4
3
2
1
117
Aug 11
Apr 11
Dec 10
Aug 10
Apr 10
Dec 09
Aug 09
Apr 09
Jan 09
0
Normalized Search Frequency
6
7
Google Search Frequency
for Illegal Immigrants and Health Care
Figure 4.2: Google search frequency for the terms “election” and “health care”
118
Figure 4.3: Search Frequency for Austerity Related Terms Over Time.
119
Figure 4.4: Tea Party Events Plotted Over Time.
120
Figure 4.5: Effect of Median Household Income on Tea Party Mobilization
121
Figure 4.6: Effect of Immigration on Tea Party Mobilization
122
Figure 4.7: Effect of Percent Black on Tea Party Mobilization.
123
Figure 4.8: Effect of Percent McCain on Tea Party Mobilization
124
Chapter 5: Searching for the Tea Party’s Origins: Google Search Data as a Predictor of StateLevel Tea Party Mobilization
The findings in chapter three establish the demographic characteristics of the Tea Party
movement and highlight the attitudes of Tea Party supporters and activists toward the economy,
race and ethnicity, and cultural values. The findings show that Tea Party members generally
occupy high-status positions in society and are more likely to be white, male, conservative and
well off. However, the findings also suggest sources of threat that are likely motivating support
for the movement. Both activists and supporters expressed hostility toward blacks and
immigrants and Tea Party activists expressed fear of downward mobility and were more likely to
have been unemployed than the general public. These findings suggesting that Tea Party
members are high status individuals who are responding to perceived threats to their status are
consistent with previous work on other right-wing movements in the US and with other work,
specifically, on the Tea Party movement which has found nativism and anti-immigrant sentiment
to be strong predictors of Tea Party support. Additionally, journalistic accounts have highlighted
a fear of downward mobility among Tea Party activists. This previous research has also
suggested that economic and racial/ethnic threat are not entirely analytically separate as antiimmigrant sentiment in the Tea Party may be tied to the sense that immigrants, and particularly
illegal immigrants, are unfairly benefiting from government programs for which Tea Party
members perceive themselves as being unfairly forced to pay.
The research in chapter 4 supports this theory, showing that Tea Party events were more
likely to occur in counties with high international immigration rates during the early period of
the movement when the debate over the Obama administration’s proposed health care reform bill
125
was dominating the national political discourse. Additionally, chapter 4 demonstrated that there
is a significant interaction between the number of recent international immigrants in a state and
national health care discourse as measured by Google search and Fox News articles. In other
words, the positive effect of immigration on Tea Party mobilization was contingent on the
national discourse about health care policy. Taken together, these results provide strong evidence
of racialized support for the Tea Party movement and racialized opposition to health care reform.
Chapter 4 also demonstrated that economic threat, as measured by median income, was more
salient in the later stages of the movement when the focus was on the 2010 midterm
congressional elections. Furthermore, chapter 4 provides some evidence that this increase in the
effect of economic threat was tied to discourse about taxation and spending issues, although the
evidence for this is weaker than the evidence in support of the connection between immigration
and health care discourse.
While chapter 3 provided evidence that individual attitudes such as anti-immigrant and
anti-black sentiment as well as perceived economic threat are correlated with Tea Party support
and activism and chapter 4 demonstrated that local sources of threat are associated with Tea
Party mobilization, contingent on the national media environment, neither chapter effectively
examined the effect of pre-existing attitudes on Tea Party mobilization. While the emergence of
right-wing and conservative movements is tied to threats caused by social and economic change,
it is true that not every individual who experiences these changes will experience them as
threatening. Previous literature provides some evidence that certain stable attitudinal
characteristics, net of economic and social positions make individuals more predisposed to find
social changes threatening, particularly in the case of outgroup and cultural threat (e.g. Adorno et
126
al. 1950; Gilens 1999). Establishing effective measures of these pre-existing cultural attitudes is
important for understanding their role in right-wing mobilization.
As has been well documented, survey data is often ineffective at measuring socially
undesirable attitudes because respondents may be reluctant to express potentially socially
undesirable opinions such as those associated with racial animosity (e.g. Presser and Stinson
1998; Podsakoff, MacKenzie, Lee and Podsakoff 2003). As a result of this desirability bias, it is
often difficult to get measures of sensitive topics such as attitudes toward race, whether or not
one votes or issues related to employment and income. While, survey researchers do attempt to
measure these, often with indirect questions such as those employed in chapter 3, it is difficult to
know how effectively these items are accomplishing their task. However, there is growing
evidence that alternative sources of data – those generated by internet users in the course of their
daily activities- can effectively measure social attitudes and dispositions in ways survey data
cannot (See chapter 2 for a more complete discussion). The remainder of this chapter will use
data constructed from Google search data to predict Tea Party mobilization at the state-level. The
findings show that pre-existing animosity toward immigrants is positively and significantly
associated with Tea Party mobilization, even after controlling for other factors. Additionally,
economic threat, as measured, by Google searches for employment during the recession are
significantly and positively associated with Tea Party mobilization. Neutral immigration related
searches – those that do not express animosity – are not significantly associated with Tea Party
mobilization, nor are searches for other cultural issues relevant to conservatism.
Google Search Data, Economic and Cultural Threat, and Tea Party Mobilization
127
In this chapter, I utilize Google search data to construct several measures of social
attitudes and behavior. I draw on the theoretical approaches described in previous chapters to
identify key words relevant to the forms of threat shown to be instrumental in Tea Party
mobilization. First, I construct two measures of attitudes toward immigrants and immigration
based on the search terms “immigrants” and “illegals.” The metric based on the term
“immigrants” is meant to measure public interest in immigration using a neutral term. The term
“immigrant” does not generally have a negative connotation and is not considered a slur or
offensive term against immigrants. High search frequencies for this term should signal public
interest in issues relating to immigration, but not anti-immigrant sentiment. In other words, being
interested in immigration should make one more likely to search for this term, but harboring
animosity toward immigrants should not.
A second immigration related metric based on the term “illegals” is also employed.
Unlike the term immigrants, the term “illegals” is often considered pejorative and offensive and
is favored by conservative opponents of immigration.1 The pejorative nature of this term should
indicate that it is more likely to be searched for by those who oppose immigration or harbor
animosity toward immigrants. As such, higher search frequencies for this term should indicate
higher levels of anti-immigrant sentiment2. The time frame used for each immigration related
term is January 2007 through December 2008. This time frame was chosen in order to establish
an estimate of pre-existing attitudes toward immigration from the time period just before the Tea
1
For example, the Associated Press and other news organizations have ceased using the related term “illegal
immigrant” and never used “illegals” to describe people: http://blog.ap.org/2013/04/02/illegal-immigrant-nomore/. Interestingly, Fox News, a popular Tea Party outlet, does use the word “illegals” as a noun to refer to
immigrants: eg. http://www.foxnews.com/politics/2011/11/23/gingrich-on-top-gop-polls-takes-big-riskarticulating-illegal-immigration/#ixzz1eXVVkuEU. Additionally, the Tea Party Patriots official website
(teapartypatriots.org) uses the word “illegals” or “illegal immigrants/aliens” 52 times (as of 3/29/2014).
2
Appendix 5A shows related search terms for both ‘immigrants’ and ‘illegals.’
128
Party movement began. The years 2009 and beyond are excluded so as to eliminate the
possibility that the Tea Party movement itself is influencing or driving the results.
Third, I construct a measure to capture economic stress that could lead to the perception
of economic threat. This measure is constructed from search frequency data for the term “jobs.”
The term “jobs” is searched for primarily by individuals who are seeking employment. As such,
higher search frequencies for this term should indicate a larger number of people who are
unemployed or underemployed in a particular area and as such should provide a good measure of
economic threat. Table 1 in appendix 5A shows related searches for this term. Most of the related
searches include references to specific types of employment or websites designed to aid job
seekers. The time frame used for this term is January 2009 through August 2011 – the same time
frame over which data on Tea Party events exists.
Finally, I utilize a fourth measure that is intended to tap into expressions of social
conservatism. This measure is constructed using the term “guns” and is designed to capture
interest in purchasing firearms and gun culture.3 Previous research has indicated that gun
ownership and support for the liberalization of gun laws is associated with a range of
conservative beliefs (Jiobu and Curry 2001; O’Brien et al. 2013). Gun owners and opponents of
gun control legislation tend to be less trusting of the federal government, more culturally
conservative, and express greater animosity and fear toward racial minorities. As such, areas
with greater search frequencies for the term ‘guns’ should have higher levels of these attitudinal
characteristics. The validity of this measure is supported by the fact that it correlated with
McCain’s 2008 vote share at 0.82. The time frame for this measure, like, the immigration
3
Several terms were actively excluded in this search to filter out potential noise. Specifically, the words “toy” and
“roses” were filtered out to block searches for toy guns and the popular music group, “Guns N’ Roses.”
129
measures is January 2007 through December, 2008. Pre-existing interest in guns is measured to
avoid potential feedback from the Tea Party movement.
The data obtained to construct these measures comes from Google Trends, a service
provided by Google, Inc to provide aggregate data on the search frequency associated with
selected terms. There are a number of technical challenges involved in obtaining this data in a
form that is useful for the style of analysis employed in this chapter. Those challenges and my
solutions to them are discussed in more detail in chapter 2.
Table 5.1 shows basic descriptive statistics for all four Google search frequency measures.
[TABLE 5.1 ABOUT HERE]
Hypotheses regarding Google search data:
H1: Search frequency estimates for the word ‘immigrants’ will not be significantly correlated
with Tea Party mobilization as this neutral term does not measure anti-immigrant sentiment.
H2: Search frequency estimates for the word ‘illegals’ will be significantly and positively
correlated with Tea Party mobilization as it measures anti-immigrant sentiment.
H3: Search frequency for ‘jobs’ will be positively and significantly correlated with Tea Party
mobilization as it measures economic stress that is likely to lead to the perception of economic
threat.
130
H4: Search frequency for the term ‘guns’ may be positively correlated with Tea Party
mobilization as it measures socially conservative attitudes.
Analysis
In order to assess the effect of the Google search frequency measures on state-level Tea Party
mobilization, negative binomial regression analysis is used to regress a state-level measure of
Tea Party mobilization on independent variables. Negative binomial regression analysis is a
method for modeling count variables with overdispersion (Long 1997). Likelihood ratio tests,
show that the dependent variable is overdispersed making the negative binomial regression
model appropriate. Table 5.2 shows descriptive statistics for the dependent variable the other
independent variables in the model.
[TABLE 5.2 ABOUT HERE]
The dependent variable is a measure of the number of Tea Party events that occurred in each
state between January 2009 and August 2011. This measure was constructed by webscraping the
Tea Party Patriots online event calendar and is discussed in greater detail in chapter 2. State
population is the US census 2009 population estimate for each state. The %White variable is the
US census 2009 estimate for the percentage of each state’s population identifying as white.
Median wage is taken from the 2008 estimate of the Bureau of Labor Statistics (BLS).
Additionally, the unemployment change measure consists of the 2007 state unemployment rate
subtracted from the 2009 rate – both come from the BLS. This measure is intended to capture the
extent to which the economic crisis damaged each state’s economy. The %McCain measure,
taken from the Federal Election Commission, is the percentage of votes received by John
McCain in each state during the 2008 Presidential election. This variable is intended to measure
131
conservatism and Republican affiliation. Finally, the immigrants measure, taken from the US
Census bureau is a measure of the number of foreign immigrants received by each state between
2000 and 2009.
RESULTS
Table 5.3 shows the results of negative binomial regression analyses predicting Tea Party
mobilization at the state-level using Google search frequencies for immigration related terms,
“immigrants” and “illegals” as the key predictor variables.
[TABLE 5.3 ABOUT HERE]
Columns 1 and 2 show that the effect of search frequency for the word “immigrants” has a small
negative and highly insignificant effect on Tea Party mobilization at the state level. However,
columns 3 and 4 show a positive, significant effect of search frequency for the word “illegals” on
Tea Party mobilization at the state level (p<.05). This result is consistent with expectations as the
word “illegals” is generally considered to be pejorative and, as noted in previous research (e.g.
Williamson et al. 2011), the issue of illegality is closely tied to Tea Party animosity toward
immigrants. Thus, these results provide evidence that pre-existing anti-immigrant sentiment, as
measured by search frequency for the word “illegals” is associated with Tea Party mobilization
even controlling for other factors. Other significant independent variables in the full models
include median wage and the measure of foreign immigrants. Median wage has a positive effect
(p<.05) across both models and the observed immigration measure has a negative, significant
effect across both models (p<.001).
The magnitude of the effect of the Google search rate for “illegals” is substantial. The
estimated rate parameter, or expected number of events, for the full model when the search
132
frequency for “illegals” is held one standard deviation below its mean and all other items held at
their mean is 54.28. When the value of the “illegals” search item is increased to one standard
deviation above its mean the estimated rate parameter increases to 85.35. Thus, the effect of preexisting anti-immigrant sentiment has a substantial impact on expected Tea Party mobilization,
even after controlling for actual state-level immigration. In fact, the Google search antiimmigrant sentiment measure is clearly picking up something distinct from immigration as
“illegals” is correlated with immigrants as a percent of population at only .3317 and with the raw
number of new immigrants at only .1814.
To provide an indication of the relative effect of the “illegals” measure compared to other
independent variables, a change in the value of the median wage from one standard deviation
below its mean to one standard deviation above its mean, holding all else at its mean, is
associated with a change in estimated rate of 53.34 to 86.836. State median wage and the antiimmigrant sentiment measure, therefore seem to have very similar overall effect sizes on Tea
Party event count at the state-level. Additionally, a similar change from the 25th to 75th percentile
in the immigration measure is associated with a change in the estimated rate of Tea Party events
from 111.63 to 65.2094.
Table 5.4 shows the results of the negative binomial regression analyses predicting statelevel Tea Party mobilization using Google search frequencies for the words “jobs” and “guns.”
In this model, “jobs” is expected to pick up economic threat, essentially measuring the extent to
which people in a given state were searching for jobs during the recession and “guns” is intended
to tap into some pre-existing conservative cultural sentiment (see O’Brien et al. 2013).
[TABLE 5.4 ABOUT HERE]
4
A slightly different range was used to demonstrate the magnitude of the immigration measure, because due to its
right skew a +/- 1sd change interprets the effect outside the range of the data.
133
Columns 1 and 2 show the effect of search frequency for the word “jobs” in both a base model,
including state population as the only other variable, and the effect in the full model. There is a
positive, significant effect of jobs in both models (p<.01). This indicates that economic threat, as
measured by search frequency for “jobs”, has a significant effect on Tea Party mobilization even
net of other factors such as observed unemployment. The magnitude of this effect is substantial
as well. Holding all other covariates at their means, when the “jobs” measure is held one
standard deviation below the mean, the estimated rate is 52.38 and when “jobs” is increased to
one standard deviation above the mean, the estimated rate increases to 87.801. Thus, even
controlling for observed unemployment the measure of economic threat captured by Google
search is a significant and substantively important predictor of Tea Party mobilization at the state
level. Other independent variables that are significant in the full model include median wage and
the immigration measure. Median wage has a positive and significant effect on Tea Party
mobilization (p<.05) and immigration has a negative and significant effect on mobilization at the
state level (p<.01).
Columns 3 and 4 show the results of both the base and full models for Google search
frequency for the word “guns.” Both are insignificant, indicating that cultural conservatism, as
measured by searches for “guns” has no effect, either by itself or net of other factors, on statelevel Tea Party mobilization. Interestingly, as noted above, search frequency for “guns” has a
correlation 0.82 with McCain’s 2008 vote share, indicating that it is a good measure of
conservatism. However, neither conservatism, as measured by republican vote share or guns
seems to be an important predictor for local mobilization. This finding is confirmed by all
models in chapter 4.
CONCLUSIONS
134
The results of this analysis have shown that the Google search frequency measures for ‘illegals’
and ‘jobs’ have a positive effect on state-level Tea Party mobilization net of other factors. This is
likely due to the fact that these measures are picking up attitudinal and behavioral conditions
within states that are correlated with the experience of racial and economic threat. Specifically
search frequency for the word “illegals” is likely associated with animosity toward immigrants.
This is a term that is frequently employed by those who are hostile toward immigrants and is
avoided by most mainstream news outlets. Chapter 4 showed that local conditions, in this case
high rates of immigration, are associated with the experience of threat if these conditions are
made salient by the national media discourse. However, the findings of this chapter demonstrate
that pre-existing attitudes toward immigration are also associated with mobilization net of local
conditions and the media.
As noted, the jobs measure is also significant and has a positive effect on Tea Party
mobilization. This measure is likely picking up economic threat by providing a gauge of the
extent to which individuals are actively seeking jobs. The related search terms for this word,
provided in appendix 5A, clearly show that the vast majority of searches for this term are related
to job seeking behavior. Thus, the search volume for this term seems to be acting as an aggregate
measure of job searching behavior which is associated with high unemployment and
underemployment. It is interesting to note that the official unemployment measure is not
significant in this model. It is possible that the Google search measure outperforms the official
measure, due to several potential shortcomings of official unemployment statistics. Namely,
official unemployment statistics do not count workers who feel that they are underemployed
(either working part time or working in a job they deem unsatisfactory), despite the fact that
underemployment may be a significant source of economic stress. Additionally, the Bureau of
135
Labor Statistics indicates that unemployment statistics, specifically, do not count unemployed
individuals who use the internet to seek jobs (unless the job seeker directly contacts the potential
employer).5 The omission of these individuals also has the potential to leave out a considerable
number of job seekers, given the increasing importance ubiquity of the internet in the daily lives
of Americans.
Unlike the search volumes for the term “illegals,” the term “immigrants” did not have a
statistically significant effect on Tea Party mobilization. Given that the term “immigrants,” as
evidenced by table 3 in Appendix 5A, does not seem to be picking up on hostility toward
immigrants or anti-immigrant sentiment, this is not surprising. Indeed, tables 1 and 2 in
Appendix 5B show that the states with high values for searches on “immigrants” are very
different from those with high values for “illegals.” High search volumes for “illegals” tend to be
focused on southwestern and southern states whereas searches for “immigrants” are highest in
the northeast, Midwest and Hawaii.
It is also interesting that search volumes for the term “guns” did not have a significant
effect on Tea Party mobilization. It should be noted, however, that this is not because the
variable is not effectively capturing aspects of conservative culture, as the item correlates very
highly with McCain’s 2008 vote share. However, neither McCain vote share, nor the guns
measure have a significant effect on Tea Party mobilization. This finding is also consistent with
the findings from chapter 4, which indicate that more conservative states and counties do not
seem to have more Tea Party events net of other factors. This finding is consistent with theories
that emphasize threat as the catalyst driving right-wing mobilization. While right-wing
movements do draw on conservative constituencies, high concentrations of conservatives in an
5
http://www.bls.gov/cps/cps_htgm.htm#unemployed
136
area do not necessarily indicate high levels of threat. In fact, homogeneously conservative areas
may have fewer factors associate with threat, like high levels of immigration or racial diversity.
The findings of this chapter are substantively interesting, particularly in that they show
that pre-existing levels of anti-immigrant sentiment are associated with higher levels of Tea
Party mobilization, event net of observed levels of immigration into the state. The findings also
provide strong evidence that economic threat motivated by unemployment and job related
insecurity is a major driver of Tea Party activism. However, this chapter also contributes to the
evidence that is beginning to accumulate in the literature showing that internet search traffic can
be a an effective tool for social scientists interested in measuring social attitudes and behaviors
and that this data can provide an alternative to survey data in some instances. Search data has a
number of potential advantages over survey data. First, search data is unlikely to suffer from
desirability bias or self-censoring. Because searches take place in private and the non-aggregate
data is not accessible to anyone else, individuals are unlikely to withhold potentially offensive or
undesirable opinions in their search behavior. Additionally, although search engines are not
designed specifically to measure attitudes or opinions, the interests of individuals, and the
language they choose to use to express those interests, is made available in this data. Aggregate
search data is also potentially more flexible than survey data in the sense that researchers are not
restricted to the questions available in existing surveys.
At the same time, there are a number of potential limitations to the use of search data.
First, as demonstrated in chapter 2, it is not particularly easy to obtain this data at the level of
small geographical areas and for many geographical areas in the US smaller than states, it may
be impossible. Furthermore, at the present time there is no obvious method available for
evaluating the appropriateness of terms as a measure for certain attitudes or behaviors besides
137
drawing on theory and the related search terms provided by the service. Finally, the availability
of this data is completely dependent on the willingness of the companies that manage search
engines to continue providing it. However, despite these potentially shortcomings this data
clearly shows promise as a useful measure for social scientists and warrants continued
exploration.
138
Table 5.1: Description statistics for each of the Google search volume measures
Variable
Mean
Minimum
Maximum
Time Frame
Immigrants
-0.03
-1.75
2.2
Jan 2007 - Dec 2008
Illegals
-0.01
-1.68
4.48
Jan 2007 - Dec 2008
Jobs
80
53
100
Jan 2009 - Aug 2011
Guns
67.92
46
100
Jan 2007 - Dec 2008
139
Table 5.2: Descriptive statistics for other variables in used in state-level negative binomial
regression models.
Variable
Events
Mean
StdDev
Minimum
Maximum
103.78
126.86
0
748
6128.14
6804.7
544.27
36961.66
%White
81.74
11.89
30.22
96.25
Median Wage
Unemployment
Change
42.08
6.5
32.82
60.29
4.1
1.43
1.2
7
%McCain
47.84
9.49
26.4
65.6
Immigration Rate
178.4
316.48
3.04
1816.63
Population
140
Table 5.3: Negative binomial regression results showing effect of independent variables on statelevel Tea Party mobilization.
Model 1
Model 2
Model 3
Model 4
Population
0.0001***
0.00005***
0.0001***
0.0002***
(6.474)
(5.391)
(6.565)
(5.810)
Immigrants
-0.001
(-0.008)
-0.004
(-0.032)
Illegals
0.228*
(1.985)
0.230*
(2.215)
%White
0.005
(0.634)
0.004
(0.556)
Median Wage
0.040*
(2.391)
0.037*
(2.416)
0.095
(1.377)
0.046
(0.664)
0.019
(1.324)
0.008
(0.659)
Unemp
Change
%McCain
Vote
Immigration
Constant
Ln(alpha)
_cons
-0.003***
(-3.419)
-0.003***
(-3.853)
3.507***
(22.629)
-0.09
(-0.064)
3.532***
(24.144)
-0.870***
(-4.178)
-1.221***
(-5.531)
-0.956***
(-4.541)
Notes: z statistics in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001
141
0.819
(0.595)
-1.330***
(-5.960)
Table 5.4: Negative binomial regression results showing effect of independent variables on statelevel Tea Party mobilization.
Model 1
Model 2
Model 3
Model 4
Population
0.0001***
0.00004***
0.0001***
0.0002***
(6.981)
(5.155)
(6.221)
(5.466)
Jobs
0.028**
(2.890)
0.027**
(2.658)
Guns
0.002
(0.182)
0.009
(0.682)
%White
0.011
(1.429)
0.004
(0.549)
Median Wage
0.041**
(2.674)
0.045*
(2.489)
0.033
(0.481)
0.097
(1.417)
0.005
(0.421)
0.012
(0.783)
Unemp
Change
%McCain
Vote
Immigration
Constant
Ln(alpha)
_cons
-0.003**
(-3.092)
1.337
(1.760)
-1.030***
(-4.815)
-0.003***
(-3.514)
-1.833
(-1.270)
3.388***
(5.068)
-1.367***
(-6.044)
-0.870***
(-4.181)
Notes: z statistics in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001
142
-0.539
(-0.345)
-1.230***
(-5.571)
Appendix 5A
Table 1: Related terms for ‘jobs’
hospital jobs
100
usa jobs
80
craigslist
70
government jobs
60
florida jobs
55
online jobs
55
part time jobs
55
steve jobs
45
monster
45
monster jobs
45
teaching jobs
40
va jobs
40
yahoo jobs
40
nursing jobs
40
security jobs
40
google jobs
35
summer jobs
35
chicago jobs
35
indeed
30
indeed jobs
30
federal jobs
30
atlanta jobs
25
indeed.com jobs
25
indeed.com
25
blow jobs
25
hot jobs
entry level jobs
ups
walmart
ups jobs
construction jobs
job search
walmart jobs
marketing jobs
utah jobs
wisconsin jobs
orlando jobs
las vegas jobs
work from home
local jobs
monster.com
jobs in atlanta
accounting jobs
target jobs
career builder
jobs in nyc
jobs in houston
cna jobs
alaska jobs
apple jobs
143
20
20
20
20
20
20
20
20
20
15
15
15
15
15
15
15
15
15
15
10
10
10
10
10
10
Table 2: Related terms for “illegals”
illegal immigrants
100
illegals social security
85
illegal immigration
80
illegal aliens
55
144
Table 3: Related terms for ‘immigrants’
the immigrants
100
immigration
65
us immigrants
35
immigrant
35
new immigrants
25
american immigrants
25
immigrants to america
20
immigrants in america
20
irish immigrants
20
immigrants in us
20
ellis island immigrants
15
ellis island
15
mexican immigrants
15
chinese immigrants
15
u.s. immigrants
15
german immigrants
10
italian immigrants
10
us immigration
10
famous immigrants
10
immigrants jobs
10
legal immigrants
10
english immigrants
10
history of immigrants
10
the new immigrants
10
immigrants in usa
5
african immigrants
immigrants rights
undocumented immigrants
early immigrants
immagrants
russian immigrants
asian immigrants
polish immigrants
european immigrants
hispanic immigrants
japanese immigrants
pictures of immigrants
city of immigrants
jewish immigrants
spanish immigrants
social security immigrants
19th century immigrants
old immigrants
jobs for immigrants
mexican immigration
illigal immigrants
indian immigrants
irish immigration
immigration statistics
french immigrants
145
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
Table 4: Related terms for ‘guns.’
gun
guns for sale
air guns
hand guns
guns america
bb guns
used guns
buy guns
machine guns
air soft guns
guns and ammo
guns online
impact guns
glock guns
glock
remington guns
handguns
remington
ruger
shot guns
ruger guns
world guns
nerf guns
young guns
pellet guns
100
40
30
25
20
20
15
15
15
15
15
10
10
10
10
10
10
10
10
10
10
10
10
10
10
big guns
taurus
taurus guns
cheap guns
shotguns
hunting guns
winchester
winchester guns
colt guns
paint guns
beretta
springfield guns
browning guns
beretta guns
smith and wesson
old guns
stun guns
guns germs steel
military guns
walmart guns
guns magazine
nail guns
gallery of guns
kimber guns
obama guns
146
10
10
10
10
10
10
10
10
10
10
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
Appendix 5B
Table 1: State rankings for ‘Immigrants’ value
Rank State
‘Immigrants’
1
New York
2.196285
2
Hawaii
2.120757
3
Iowa
1.65582
4
Illinois
1.407209
5
Connecticut
1.327975
6
Minnesota
1.193449
7
Wisconsin
1.066358
8
Rhode Island
0.946942
9
Washington
0.886501
10
Vermont
0.739083
11
Indiana
0.680968
12
Massachusetts 0.671131
13
Delaware
0.666746
14
Maryland
0.595087
15
Pennsylvania
0.594215
16
Arizona
0.554956
17
Maine
0.533049
18
California
0.436008
19
Nebraska
0.376015
20
New Jersey
0.288284
21
Colorado
0.264367
22
Kentucky
0.146724
23
South Carolina 0.011922
24
Kansas
-0.07579
25
Idaho
-0.0775
Rank
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
147
State
Oregon
Texas
North Carolina
Michigan
Ohio
Montana
Georgia
North Dakota
New Hampshire
Utah
West Virginia
South Dakota
Missouri
Nevada
Tennessee
Florida
Arkansas
New Mexico
Alaska
Virginia
Wyoming
Mississippi
Alabama
Louisiana
Oklahoma
‘Immigrants’
-0.12067
-0.14216
-0.18169
-0.22749
-0.2298
-0.38446
-0.38545
-0.43816
-0.49682
-0.50385
-0.52151
-0.66626
-0.74074
-0.86422
-0.88225
-0.96996
-1.0824
-1.22583
-1.30528
-1.37983
-1.43608
-1.53537
-1.54712
-1.55157
-1.75046
Table 2: State rankings for ‘illegals’ value.
Rank State
‘Illegals’
1
Arizona
4.477943
2
Nevada
1.912511
3
New Mexico
1.334706
4
Tennessee
1.219407
5
Colorado
1.190426
6
Oklahoma
0.880433
7
Georgia
0.788824
8
Texas
0.773483
9
Idaho
0.694039
South
10
Carolina
0.55287
North
11
Carolina
0.419305
12
Florida
0.271125
13
Illinois
0.258686
14
Mississippi
0.244533
15
California
0.239715
16
Indiana
0.217499
17
Iowa
0.151618
18
Kansas
0.122684
19
Alabama
0.105663
20
Kentucky
0.058864
21
Arkansas
0.001354
22
Maryland
-0.00884
23
Oregon
-0.07374
24
Wyoming
-0.10548
25
Rhode Island -0.13317
Rank
26
27
28
29
30
31
32
33
34
State
New Jersey
Montana
Utah
Nebraska
Washington
Connecticut
Ohio
Pennsylvania
Massachusetts
‘Illegals’
-0.15496
-0.18313
-0.23191
-0.25394
-0.27122
-0.28313
-0.29639
-0.29777
-0.3422
35
New York
-0.38058
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Missouri
New Hampshire
Wisconsin
Louisiana
Virginia
Alaska
Michigan
South Dakota
North Dakota
Minnesota
Delaware
Hawaii
Maine
West Virginia
Vermont
-0.47315
-0.47656
-0.54202
-0.5753
-0.62687
-0.67023
-0.75073
-0.81532
-0.83922
-0.98668
-1.05696
-1.22118
-1.31434
-1.4023
-1.68462
148
Table 3: State rankings for ‘jobs’ value
Rank State
‘Jobs’
1
North Carolina 100
2
South Carolina 97
3
Georgia
96
4
Florida
95
5
Alabama
92
6
Mississippi
91
7
Maryland
91
8
Colorado
90
9
Texas
89
10
Missouri
88
11
Tennessee
87
12
Wisconsin
87
13
Arizona
86
14
Pennsylvania
86
15
Utah
85
16
Arkansas
84
17
Nevada
84
18
Ohio
84
19
Minnesota
83
20
Kentucky
82
21
Louisiana
82
22
New Mexico
82
23
Alaska
81
24
Wyoming
81
25
Idaho
81
Rank
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
149
State
Oklahoma
New Jersey
Illinois
South Dakota
Kansas
Michigan
Montana
Indiana
Washington
West Virginia
Oregon
Nebraska
Connecticut
Iowa
New Hampshire
Delaware
Maine
New York
Massachusetts
North Dakota
Rhode Island
California
Hawaii
Virginia
Vermont
‘Jobs’
81
80
79
78
78
78
76
76
76
75
75
74
74
73
73
72
71
71
70
69
69
68
68
59
53
Table 4: State rankings for ‘guns’ value
Rank State
‘Guns’
1
Alaska
100
2
Wyoming
99
3
Idaho
90
4
Montana
90
5
Utah
87
6
Arkansas
83
7
South Dakota
79
8
Kentucky
76
9
Louisiana
76
10
New Mexico
76
11
North Dakota
76
12
Oklahoma
76
13
Tennessee
76
14
Arizona
75
15
Missouri
75
16
West Virginia
75
17
Kansas
74
18
Alabama
73
19
Mississippi
72
20
Nevada
71
21
Indiana
70
22
Texas
70
23
Nebraska
69
24
North Carolina 69
25
Colorado
68
Rank
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
150
State
Georgia
Ohio
South Carolina
Maine
Florida
Pennsylvania
Washington
Michigan
Iowa
Oregon
Wisconsin
Vermont
New Hampshire
Virginia
Connecticut
Illinois
Minnesota
California
Maryland
New Jersey
Rhode Island
Hawaii
Delaware
Massachusetts
New York
‘Guns
68
68
68
67
66
64
64
63
62
62
61
59
57
56
55
55
55
52
52
52
52
51
50
46
46
Chapter 6: Conclusions and Implications
The Tea Party Movement emerged in early 2009, shortly after the election of Barack Obama. A
variety of groups claimed the Tea Party title and identity, ranging from relatively elite funding
organizations like FreedomWorks to the grassroots Tea Party Patriots who maintain a loose
network of local chapters. While the Tea Party movement maintained a consistent identity as a
conservative movement opposed to the agenda of the Democratic Party and President Obama,
the movement was characterized by a number of changes and contradictions over time. The
focus of the movement changed between early 2009 through the end of the 2010 elections as the
national political environment changed. Additionally, the group’s agenda, as expressed by elite
members and official mission statements, often conflicted with the undercurrent of racism and
cultural politics observed by outside critics.
This dissertation focuses on Tea Party mobilization at the grassroots level and contributes
to sociological understandings of how right-wing movements mobilize in a complex and
changing environment. Previous research on right-wing and conservative social movements has
identified the important role of threat in right-wing mobilization and has argued that the
widespread availability of resources for conservative movements makes resource mobilization
less problematic for these movements. However, it has largely left open the question of how
right-wing movements interact with the broader political environment. In its empirical
investigations of the Tea Party movement, this dissertation project attempts to establish a theory
of right-wing mobilization that accounts for the role of threat while also providing an explanation
for how movements interact with the broader political environment. The dissertation project also
151
makes methodological contributions and adds to our substantive understanding of an important
political social movement.
FINDINGS AND CONCLUSIONS
Summary of Findings
This dissertation begins with an examination of Tea Party supporters and activists at the
individual-level. The results of an analysis of polling data from April 2010 show that Tea Party
supporters are generally relatively high-status individuals. Tea Party supporters tend to be older,
whiter, more male and more highly educated than the general population. Tea Partiers also tend
to be more conservative and more Republican than the general population. In addition to these
general characteristics, Tea Party supporters tend to hold views that are more hostile to racial and
ethnic minorities than non-Tea Party supporters. Tea Party activists are similar in these respects
to Tea Party supporters, however they tend to be even more conservative than supporters and are
more likely to express attitudes consistent with a feeling of economic threat – they are more
likely to view their taxes as unfair and to fear falling in economic class. Tea Party activists are
also more likely to be unemployed. On the whole, Tea Party members are shown to be relatively
high-status conservatives who express hostility toward ethnic and racial minorities and who
express concern over their economic status. These findings are largely consistent with previous
literature on right-wing and conservative social movements that have found the constituents of
these movements to have similar characteristics.
Additionally, the first empirical chapter shows that the composition of the movement’s
base of supporters changed between early 2010 and 2011. During this time period, the
percentage of the population expressing support for the movement declined and supporters of the
152
movement became more homogenous, more conservative and more partisan in their support for
the Republican Party. Although, it is not possible to analyze the individual-level correlates of
actual participation in the movement over time, these findings suggest that the Tea Party had
changed over the course of the protest cycle as the political context in which it operated changed.
Although the findings of this first empirical chapter are suggestive they leave important
questions unanswered – namely, under what circumstances do social changes produce
perceptions of threat that lead to right-wing mobilization.
The second empirical chapter directly explores Tea Party mobilization using a data set of
all officially recorded Tea Party Patriot events taking place between January 2009 and August
2011. Consistent with previous literature, it finds that local conditions associated with potential
racial and economic threats to the status of white conservatives were predictive of Tea Party
event counts at the state and county levels. However, it also found that the conditions associated
with Tea Party mobilization shifted over time as the social and political context in which the
movement operated changed.
On the county-level it finds that during the early stages of the movement higher rates of
immigration were an important predictor of Tea Party mobilization. This finding is consistent
with previous research which has found that changes in the racial and ethnic composition of local
areas tends to be associated with an increased sense of threat among whites, particularly
conservative whites. Additionally, other research on the Tea Party as well as evidence presented
in this chapter indicates that the Tea Party movement was particularly sensitive to the potential
threat posed by immigration and tended to associate the threat of immigration with welfare
spending – particularly healthcare reform. These findings suggest that the health care reform
debate made the threat of immigration more salient and drove Tea Party mobilization in areas
153
with higher immigration rates. This finding is confirmed in the state-level analysis which finds
that state-level immigration has a positive and significant interaction with national discourse on
health care reform.
The second empirical chapter also found that local conditions associated with economic
threat were significant predictors of Tea Party mobilization. However, like immigration, the
effect of some economic conditions were not consistent over time. The county-level analysis
indicated that median income had a larger effect during the second time period when the 2010
midterm congressional elections dominated the national discourse. This finding is consistent with
the notion that an election discourse focusing on fiscal issues relating to taxation and spending
made economic threat more salient to higher income individuals who might view government
spending and taxation as a threat to their economic status. This hypothesis, however, received
only limited support from the state-level time series cross sectional analysis, with only some
national discourse terms interacting significantly with state-level median income.
The final empirical chapter provides both methodological and theoretical contributions. It
uses a technique developed in chapter 3 for extracting state-level data on Google search
frequencies and employs this data to measure public opinion and attitudes. This data has several
potential advantages over other methods for measuring attitudes and behavior among the public,
such as surveys: it is less likely to be affected by desirability bias and it is more flexible in that
researchers are not limited to a predetermined set of survey questions. Additionally, using
methods like the one proposed in this chapter, it is possible to get precise estimates of attitudes
and behaviors at a level of geographic granularity that is often hard to achieve with surveys.
Even with large surveys, state-level estimates become problematic in smaller states.
154
Substantively, this chapter finds that pre-existing anti-immigrant sentiment and jobseeking behavior are positively and significantly associated with Tea Party mobilization. This
supports findings from previous chapters suggesting that economic and cultural/racial threat are
drivers of Tea Party mobilization. Unlike the second empirical chapter, which showed that
higher rates of immigration (combined with high-levels of healthcare discourse) were associated
with Tea Party mobilization, the findings of this chapter indicated that pre-existing attitudes net
of actual immigration rates were also associated with mobilization. Thus, there appear to be
cultural differences in attitudes toward immigration, not directly tied to recent immigration rates
that also influence Tea Party mobilization. Additionally, Google searches for the term ‘jobs’ –a
measure of job seeking behavior and economic threat- was statistically significant, while official
unemployment statistics were not. This fact speaks to potential problems with the validity of
official unemployment statistics in picking up economic threat, possibly related to the fact that
unemployment statistics are not able to detect those who are seeking jobs because they feel they
are underemployed rather than unemployed.
Theoretical Contributions
In recent decades social movement theory has generally focused on the problem of resources and
structural opportunity. Most social movements that have attracted scholarly attention since the
second half of the 20th century have been composed of or worked on behalf of marginalized
groups who had little political access and scarce access to resources. For these movements,
securing resources and opportunities were the key organizational obstacles and the primary
ingredients required for successful mobilization. This dissertation focuses on the class of social
movements that do not fit this model – the right wing movement rich in resources and with
155
sufficient status to have access to conventional political channels. The theory implied by
previous scholars who have studied similar movements is that right-wing mobilization is largely
a function of threat. When high status groups perceive a threat to their status, they mobilize to
neutralize the threat and preserve or retain their status.
In many ways the Tea Party movement fits this model. The findings of the first empirical
chapter in this dissertation, as well as research by other scholars indicate that the Tea Party
movement is composed of individuals who are more likely to be white, high income, and male
than the general population. The movement also seems to be responding to threats to these
statuses as mobilization is associated with several forms of threat and individual Tea Party
activists tend to express such concerns in surveys. These findings are important in their own right
in that they confirm the continuing applicability of previous research findings and expand our
knowledge of an important social movement.
However, the findings of the previous chapters also indicate how the theory implied by
previous studies of right-wing movements can be modified and augmented. While the previous
literature establishes right-wing mobilization as a reaction against perceived threats, it leaves
open questions regarding how right-wing movements are affected by other social processes that
may affect the perception of threat. For example, the previous chapters of this dissertation have
shown that certain types of social change (such as immigration) are correlated with Tea Party
mobilization. It is easy to understand why conservative whites might find these changes
threatening – both to their economic and cultural statuses – however, high rates of immigration
in certain areas are not unique to the era in which the Tea Party mobilized. The question left
open is why do common forms potentially threatening social change sometimes result in rightwing mobilization and sometimes not.
156
The findings in the second chapter of this dissertation suggest that the national media discourse
is a crucial element in understanding the politicization and salience of potentially threatening
social changes. Focusing on the importance of media discourse in shaping political attitudes and
behavior is not unprecedented. There is a rich literature suggesting that the media can have
priming and agenda setting effects (e.g. Domke, Shah and Wackman 1998; Scheufele and
Tewksbury 2007). Additionally, Hopkins (2010) has suggested that national media discourse
about immigration has the potential to politicize the issue and fuel opposition to immigration
when it might otherwise be ignored. Although the issue of framing and interpretation has been
addressed by a small number of scholars of right-wing movements (most notably McVeigh
2009), the focus has been on messaging by movement leaders rather than political and social
processes external to the movement.
The findings in this dissertation show the local correlates of Tea Party behavior were
dependent on the content of the national media discourse. When health care reform dominated
the discourse, immigration and cultural threat were important predictors of mobilization – likely
because discourse surrounding the expansion of the welfare state made the presence of ethnic
and racial minorities more salient to conservatives who harbor highly racialized opposition to the
welfare state. Additionally, conditions associated with economic threat, such as median income,
are shown to have been more important when the 2010 congressional elections dominated the
national discourse. These findings show that in studies of right-wing movements, social changes
that can potentially create the perception of threat among high status groups should not be
considered outside the context of larger national debates.
Figure 1.1 in the introductory chapter suggests a model for thinking about right-wing
mobilization. It shows right-wing mobilization to be a function of social changes, perceived
157
threat and the public discourse. The chapters of this dissertation address each of these forms of
threat and suggest how they interact to produce right-wing mobilization. Social changes (e.g.
demographic shifts, immigration, election campaigns, and economic downturns) can produce
perceived economic threat, but must be activated by the national media discourse that politicizes
them and makes them seem salient and relevant to the status of the movement’s constituency.
Although resources and organizing capacity are necessary, these are taken as given in the case of
most right-wing movements.
Recent Developments in Non-Conservative Mobilization
As has been noted above, the underlying theory behind most recent literature on
conservative and right-wing mobilization implies that mobilization for conservative and rightwing groups is relatively unproblematic compared to movements made up of marginalized
groups. Scholars have argued that the ready availability of elite resources and access to
conventional forms of political power dramatically reduce the extent to which right-wing
movements are burdened by concerns about resource mobilization and political opportunities. As
such, conservative mobilization is more contingent on grievances and threat and can occur more
rapidly than non-conservative mobilization. However, recently literature has suggested that
recent advancements in communication technologies may be changing the dynamics of nonconservative mobilization in ways that make it more similar to right-wing mobilization.
Specifically, this literature suggests that these technologies are reducing the costs of organization
and identity formation and reducing the need for resources and formal organization. Much of this
literature has focused on the Occupy Wall Street movement and the Arab Spring Movements.
158
Benntt and Segerberg (2012) argue that the emergence of new forms of communication
technology – notably social media and social networking services – has fundamentally changed
the nature of the problems associated with mobilizing constituencies, faced by social movement
organizers. The authors contend that, before the emergence of these new technologies, social
movement organization was focused on solving collective action problems – essentially the
problems associated with getting individuals to contribute to a common cause when it might,
individually, be more rational to free-ride. Strategies to overcome collective action problems
have typically focused on building organizations, forging collective identities and marshaling
resources. Each of these processes requires that strong organizations play a central role in
coordinating action and managing solutions to collective action problems. Previous social
movement theories, particularly resource mobilization theory, have explained successful
mobilization as a function of social movement organizations’ ability to command outside
resources for their cause – an ability that is forged over time through years of effort.
While traditional social movement organizations still play an important role in many
social movements, the authors contend that emerging communications technologies have made
alternative strategies for mobilization possible. Specifically, the authors argue that many new
social movements mobilize according to what they call ‘the logic of connective action’ as
opposed to the ‘logic of collective action (see Olson 1965)” that has motivated previous social
movement theory. According to the logic of connective action, mobilization occurs through the
use of online media networks to spread and interpret collective action frames, channel resources
and construct identity and solidarity among activists – even those who are geographically distant
from the protest events. Unlike more traditional forms of mobilization, protest movements that
use the logic of collective action tend to be leaderless and lack the formal organizational
159
structure that characterizes social movement organizations. The logic of connective action allows
movements to rapidly achieve large scale mobilization without building an organizational
infrastructure.
Juris (2013) makes a similar point, arguing that social media and internet technologies
have allowed for the emergence of a logic of networking and a separate logic of aggregation. The
logic of networking is enabled by technologies like email listserves which allow for coordination
and facilitate the spread of collective action frames. In this sense, technology does much of the
work previously accomplished by social movement organizations. Additionally, other
technologies, namely social media, facilitate what Juris calls the logic of aggregation, or the
process of interaction that allows individuals and actors to build the sense of solidarity necessary
for mobilization.
Taken together, this literature suggests that the processes required for social movement
mobilization may be changing in ways that may potentially allow left-wing mobilization to more
closely resemble right-wing mobilization as the importance of resource intensive formal
organizations decline in the face of emerging communications technologies. Research on the
recent anti-capitalist Occupy Wall Street movement in the United States seems to provide some
empirical support for this idea showing that communication networks on the microblogging
service, Twitter, effectively assumed the role of traditional social movement organizations by
facilitating the acquisition of resources, coordinating action and communicating collective action
frames (Conover, Ferrara, Menczer and Flammini 2013; Conover, Davis, Ferrara, McKelvey,
Menczer and Flammini 2013). Social media networks allowed Occupy to bypass the need to
develop a formal organizational infrastructure and facilitated very rapid large-scale mobilization.
Additionally, Gaby and Caren (2012) found that online communities were instrumental in
160
organizing, communicating and recruiting members to the occupy movement in the United
States.
Other scholars have pointed to the Arab Spring protests that occurred in several Middle
East and North African (MENA) countries beginning in the late winter of 2010. Howard, Duffy,
Freelon, Hussain, Mari and Mazaid (2011) argue that social media was used to disseminate
information and spread collective action frames. As the authors note “using digital technologies,
democracy advocates created a freedom meme that took on a life of its own and spread ideas
about liberty and revolution to a surprisingly large number of people (p. 3).” The authors also
note that the online discourse seemed to precipitate protest events, as spikes in revolutionary
discourse on social media occurred shortly before each protest event.
Other scholars have explicitly challenged the resource mobilization model, arguing that
Arab Spring mobilization used social media in ways that allowed the movement to circumvent
the necessity of developing resources and formal organizations. Khondker (2011) argues that
social media filled a consciousness raising and networking role in the Arab Spring, partly
because of the lack of existing civil society and social movement organizations in the region.
Meanwhile, Eltantawy and Wiest (2011) argue that new communications technologies, such as
social media and social networking sites, are themselves a resource that allows movements to
organize and mobilize much faster than was possible through traditional movement
organizations.
The changing nature of mobilization in non-conservative movements as they adapt to a
changing technological landscape warrants continued attention and investigation moving
forward.
161
OTHER ISSUES AND QUESTIONS RELATED TO THE TEA PARTY MOVEMENT
A number of questions about the Tea Party movement, its effects on American politics, and its
place in the history of the American conservative movement are not directly addressed by this
dissertation, but research by other scholars does provide insight into these questions and how
they relate to the findings in the preceding chapters. One question relates to the role of elite
funders, corporate interests and the potential “top down” nature of the Tea Party movement as
well as the role these elite resources played in local mobilization and individual attitudes toward
the movement. Another question relates to the success of the Tea Party movement at influencing
the platform of the Republican Party and at successfully helping conservative Republican
politicians get elected to office. An emerging literature speaks directly to both of these questions.
Top down control, corporate resources, and the Tea Party
The increasing involvement of business interests in politics – particularly conservative
politics- has been well documented in recent years. O’Connor (2010) argues that over the course
of the second half of the 20th century, corporations began to refocus their charitable giving
toward foundations and organizations pursuing agendas that were consistent with the political
and economic interests of the business community. As a result of this new focus, the business
community began to redirect their contributions toward conservative foundations and think
tanks. Essentially, they were constructing the intellectual and identity-building infrastructure for
conservative political action.
Similarly, Walker (2009), describes the rise of grassroots lobbying firms that occurred in
the United States beginning in the 1970s. He argues that firms and business trade associations,
which had long engaged in lobbying efforts, recognized the need to give their campaigns the
162
appearance of widespread public support and the legitimacy associated with such support. To fill
this need, firms began to emerge, offering services for mobilizing members of the public. These
services entail identifying, funding, and providing infrastructure support to grassroots social
movements and organizations with goals consistent with the ideology of the funders. Walker
argues that the emergence and rise of these grassroots lobbying firms and their support for
certain types of participation has drowned out other forms of civic engagement in the public
sphere.
Many observers, and even some Tea Party leaders and organizations, have discussed
concerns or accusations about the “top down” nature of the Tea Party movement – the influence
exerted by wealthy donors, corporate interests and the Republican Party on the movement (e.g.
Skocpol and Williamson 2012). A number of media observers and critics of the movement have
argued that the Tea Party movement is an example of an “astroturf” movement; meaning that it is
a movement engineered and funded by wealthy elites to provide the façade of widespread public
support.
However, while it is true that considerable resources were made available to the Tea
Party, the emerging research suggests that these resources did not seem to directly steer local
mobilization. Fetner and King (2014) argue that the role of resources in the Tea Party movement
is largely consistent with how elite resources have been understood to operate in previous work
on conservative and right-wing social movements. That is, resources were made available at the
national-level to build infrastructure, allowing for mobilization and the crystallization of a
movement identity to occur much more rapidly than it would have without elite support. At the
same time, the authors argue that elite funders did not seem to directly control the local
mobilizing activity of Tea Party Patriot’s chapters. For example, the authors report that the Tea
163
Party Patriots national organization received large anonymous donations in the early days of the
movement, but these funds were distributed equally to active local chapters and were used to
construct the online network linking local chapters.
Additionally, Fetner and King, as well as other scholars, have pointed to the importance
of elites attempting to influence media discourse to boost Tea Party attendance and actively
shape the Tea Party identity. While numerous scholars have pointed to the relationship between
the Tea Party and Fox News, Skocpol and Williamson (2012) have also argued that conservative
talk radio as well as conservative print outlets and blogs were instrumental in rallying Tea Party
support. The effect of elite attempts to steer the direction and focus of local mobilization
activities (as opposed to simply rallying support) through manipulation of the media is
potentially a fruitful avenue of further research.
Effect of Tea Party on electoral outcomes
Other questions about the Tea Party movement not directly addressed in this dissertation relate to
the movement’s impact on the Republican Party’s electoral success in the elections that followed
the rise of the movement. A considerable literature addressing this topic has begun to emerge in
political science. Karpowitz, Monson, Patterson and Pope (2011) conducted an analysis of the
Tea Party movement’s effect on the 2010 congressional elections, by looking at the effect of
endorsements by various Tea Party groups on electoral outcomes. The findings indicate that no
Tea Party groups’ endorsements had a significant effect on election results, except those of
FreedomWorks.
The results highlight the conflicted relationship that exists between the Tea Party and the
Republican Party. The authors argue that, despite the Tea Party’s affinity toward the Republican
164
Party, a tension exists between the movement and Republican office holders and candidates who
were viewed as being insufficiently supportive of the movement’s agenda. Because of this
tension, many Tea Party groups endorsed more extreme candidates who had little chance of
winning in general elections. FreedomWorks on the other hand, being a group with closer ties to
the Republican Party and elite donors, backed more mainstream candidates and provided
substantial financial support along with their donations. Given that these groups pursued such
different endorsement strategies, the authors conclude that it is difficult to establish any causal
effects regarding endorsement and election outcomes.
Rather than focusing on endorsements, other scholars have focused on the effect of
protest events. Madestam, Shoag, Veuger and Yanagizawa-Drott (2013) take an ingenious
approach to estimating the effect of Tea Party protest events on election and policy outcomes.
Given that it is likely that unobserved policy preferences influence Tea Party mobilization,
election outcomes and policy changes, they use an instrumental variable approach to estimate the
effect of mobilization, using rainfall as the instrument. The findings show that higher levels of
mobilization during the 2009 “Tax Day” protests increased Republican turnout during the 2010
elections and boosted Republican candidates. Additionally, the authors find that mobilization
affected policy by influencing the voting patterns of incumbent representatives.
Bailey, Mummolo and Noel (2012) find that the presence of Tea Party activists is a
significant predictor of higher Republican vote shares in the 2010 election. The authors also find
that higher numbers of Tea Party activists, at the district-level, increase the likelihood that a
member of Congress will vote with the Tea Party position on any given bill. Confirming the
findings of Karpowitz et al. (2011), the authors also find that endorsement by FreedomWorks
was positively and significantly associated with the vote share of endorsed candidates. However,
165
like Karpowitz et al, they are skeptical of a causal effect, given that FreedomWorks was strategic
in only supporting candidates already deemed likely to win.
Taken together the existing research that has been conducted on the Tea Party
movement’s effect on electoral politics indicates that Tea Party mobilization and grassroots
activism was successful in boosting the prospects of Republican candidates and at pushing
candidates and incumbents to the right on policy issues. Group endorsements, however, seem to
have little if any discernable effect on electoral outcomes.
Conclusion
This dissertation project has used the emergence of the Tea Party movement as an opportunity to
improve sociological understandings of right-wing mobilization processes. It has found that Tea
Party mobilization is, in many ways, consistent with previous research on right-wing movements
in that the Tea Party is a response to threats to the status of conservative, financially well off
whites. However, this project also emphasizes the importance of national media discourse in
politicizing potential forms of threat and making them salient. In addition to its theoretical
contributions, this dissertation project also provides a methodological contribution by
demonstrating a technique for producing valid and useful state-level estimates of attitudes and
behavior from Google search data. Future research questions may involve the role of elite
funders in strategically using the influence of the media to direct and manipulate Tea Party
mobilization.
166
References:
Abramowitz, Alan. 2011. “Partisan polarization and the rise of the Tea Party
Movement”. In annual meeting of the American Political Science Association (pp. 1-4).
Adorno, Theodor. W., Else Frenkel-Brunswik, Daniel J. Levinson, and R. Nevitt Sanford.
1950. The authoritarian personality. Oxford, England: Harpers.
Agarwal, Sheetal D., W. Lance Bennett, Courtney N. Johnson, and Shawn Walker. 2014. "A
Model of Crowd Enabled Organization: Theory and Methods for Understanding the Role
of Twitter in the Occupy Protests." International Journal of Communication 8: 27.
Armey, Richard, Jack Kemp and C. Boyden Gray. 2004. “Citizens for a Sound Economy
(CSE) and Empower America Merge to Form FreedomWorks”. Retrieved May 16, 2012
from http://web.archive.org/web/20040725031033/http://www.freedomw
orks.org/release.php
Armey, Richard K and Matt Kibbe. 2009. Give Us Liberty: A Tea Party Manifesto. New
York: HarperCollins.
Ashbee, Edward. 2011. "Bewitched-The Tea Party Movement: Ideas, Interests and
Institutions." Political Quarterly 82:157-164.
Associated Press. 2010. “Tea Party convention loses two Republican lawmakers over
ethics concerns.” Washington Post, January 29. Retrieved May 16, 2012 from
http://washingtonpost.com/wp-dyn/content/article/2010/01/28//AR2010012803565.html
Asur, Sitaram, and Bernardo A. Huberman. 2010. "Predicting the future with social media."
In Web Intelligence and Intelligent Agent Technology (WI-IAT), 2010 IEEE/WIC/ACM
International Conference on, vol. 1, pp. 492-499. IEEE.
Bailey, Michael A., Jonathan Mummolo, and Hans Noel. 2012. "Tea Party Influence: A Story of
167
Activists and Elites." American Politics Research 40, no. 5: 769-804.
Barberá, Pablo, and Gonzalo Rivero. 2010. "Understanding the political representativeness of
Twitter users." Unpublished Manuscript.
Bartholomew, David, Fiona Steele, Irini Moustaki and Jane Galbraith. 2008. Analysis of
Multivariate Social Science Data. Boca Raton, FL: Taylor and Francis
Baum, Matthew and Tim Groeling. 2008. "New Media and the Polarization of American
Political Discourse." Political Communication 25:345-365.
Bennett, Lance and Alexandra Segerberg. 2012. “The Logic of Connective Action.” Information,
Communication and Society 15(5): 739-768
Betz, Hans-George. 1993a. "The New Politics of Resentment: Radical Right-Wing
Populist Parties in Western Europe." Comparative Politics 25(4):413-27.
Betz, Hans-George. 1993b. "The Two Faces of Radical Right-Wing Populism in Western
Europe." The Review of Politics 55:663-685.
Blalock, Hubert M. 1967. Toward a Theory of Minority Group Relations. New York: Wiley
Blee, Kathleen M. 2002. Inside organized racism: women in the hate movement.
Berkeley: University of California Press.
Blee, Kathleen M. and Ashley Currier. 2006. "How Local Social Movement Groups
Handle a Presidential Election." Qualitative Sociology 29:261-280.
Bollen, Johan, Huina Mao, and Xiaojun Zeng. 2011. "Twitter mood predicts the stock
market." Journal of Computational Science 2, no. 1: 1-8.
Brownstein, John S., Clark C. Freifeld, and Lawrence C. Madoff. 2009. "Digital disease
detection—harnessing the Web for public health surveillance." New England Journal of
Medicine 360, no. 21: 2153-2157.
168
Brustein, William. 1996. The logic of evil : the social origins of the Nazi Party, 19251933. New Haven: Yale University Press.
Burghart, Devin and Leonard Zeskind. 2010. "Tea Party Nationalism." Retrieved May 12, 2012
From http://www.irehr.org/issue-areas/tea-party-nationalism.
Burns, Alex and Ben Eltham. 2009. “Twitter free Iran: an evaluation of Twitter’s role in public
diplomacy and information operations in Iran’s 2009 election crisis.” In Communications
Policy Research Forum 2009. University of Technology, Sydney
Carneiro, Herman Anthony, and Eleftherios Mylonakis. 2009. "Google trends: a web-based tool
for real-time surveillance of disease outbreaks." Clinical infectious diseases 49, no. 10:
1557-1564.
Choi, Hyunyoung and Val Harian. 2012. “Predicting the Present with Google Trends.” Economic
Record 88:2-9
Conover, Michael, Clayton Davis, Emilio Ferrara, Karissa McKelvey, Filippo Menczer, and
Alessandro Fammini. 2013. “The Geospatial Characteristics of Social Movement
Communication Network.” PLoS ONE 8(3): e55957. Doi:10.1371/journal.pone.0055957
Conover, Michael, Emilio Ferrara, Filippo Menczer and Alessandro Flammini. 2013. “The
Digital Evolution of Occupy Wall Street” PloS ONE 8(5): e64679.
Doi:10.1371/journal.pone.0064679
Courser, Zachary. 2012. "The Tea 'Party' as a Conservative Social movement." Society
49:43-54.
Crepaz, Markus ML, and Regan Damron. 2009. "Constructing Tolerance: How the Welfare State
Shapes Attitudes About Immigrants." Comparative Political Studies 42(3): 437-463.
Davies, James C. 1962. "Toward a Theory of Revolution." American Sociological
169
Review 27(1):5-19.
Davis, Julie. 2010. “Tea party seeks Capitol clout after election gains”, Bloomberg
Businessweek, November 17 Retrieved May 16, 2012 from
http://www.businessweek.com/ap/financialnews/D9JHUBAO0.htm
DiGrazia, Joseph. 2014. “Individual Protest Participation in the United States: Conventional and
Unconventional Activism.” Social Science Quarterly 95:111-131
DiGrazia, Joseph, Karissa McKelvey, Johan Bollen and Fabio Rojas. 2013. “More Tweets, More
Votes: Social Media as a Quantitative Indicator of Political Behavior.” PLOS ONE 8(11):
e79449. doi:10.1371/journal.pone.0079449
Dodds, Peter Sheridan, Kameron Decker Harris, Isabel M. Kloumann, Catherine A. Bliss, and
Christopher M. Danforth. 2011. "Temporal patterns of happiness and information in a
global social network: Hedonometrics and Twitter." PLOS ONE 6 (12): e26752.
Domke, David, Dhavan V. Shah, and Daniel B. Wackman. 1998. "Media priming effects:
Accessibility, association, and activation." International Journal of Public Opinion
Research 10, no. 1: 51-74.
Eltantawy, Nahed and Julie Wiest. 2011. “Social Media in the Egyptian Revolution:
Reconsidering Resource Mobilization Theory” International Journal of Communication
5: 1207-1224
Evans, Peter. 1997. “The Eclipse of the State? Reflections on Stateness in an Era of
Globalization.” World Politics Vol. 50, No.1. 62-87
Feierabend, Ivo K., and Rosalind L. Feierabend. 1966. "Aggressive Behaviors within
Polities, 1948-1962: A Cross-National Study." The Journal of Conflict Resolution
10(3):249-71.
170
Fligstein, Neil and Doug McAdam. 2011. “Toward a General Theory of Strategic Action Fields”
Sociological Theory 29:1-26
Gaby, Sarah and Neal Caren. 2012. “Ocupy Online: How Cute Old Men and Malcom X
Recruited 400,000US Users to OWS on Facebook.” Social Movement Studies 11: 367374
Gilens, Martin. 1999. Why Americans hate welfare : race, media, and the politics of
antipoverty policy. Chicago: University of Chicago Press.
Golder, Scott A., and Michael W. Macy. 2011. "Diurnal and seasonal mood vary with work,
sleep, and daylength across diverse cultures." Science 333, no. 6051: 1878-1881.
Good, Chris. 2010. Tea Party Convention Organizers Speak Out, The Atlantic, January
30. Retrieved May 16, 2012 from
http://www.theatlantic.com/politics/archive/2010/01/tea-party-convention-organizersspeak-out/35053/#
Good, C. 2010. “Is Palin's Tea Party Speech a Mistake? Tea Partiers Have Mixed
Opinions”, The Atlantic, February 4. Retrieved May 12, 2012 from
http://www.theatlantic.com/politics/archive/2010/02/is-palins-tea-party-speech-amistake-tea-partiers-have-mixed-opinions/35360/
Gould, Roger V. 1991. "Multiple Networks and Mobilization in the Paris Commune,
1871." American Sociological Review 56(6):716-29.
Gurr, Ted Robert. 1970. Why men rebel. Princeton, N.J: Princeton University Press.
Gusfield, Joseph R. 1963. Symbolic crusade; status politics and the American
temperance movement. Urbana,: University of Illinois Press.
Fetner, Tina and Brayden King. 2014. “Three-Layer Movements, Resources and the Tea Party”
171
in Understanding the Tea Party, eds. Nella Van Dyke and Divid Meyer. Burlington, VT:
Ashgate.
Hardisty, Jean V. 1999. Mobilizing resentment conservative resurgence from the John
Birch Society to the Promise Keepers. Boston: Beacon Press.
Heaney, Michael T. and Fabio Rojas. 2011. "The Partisan Dynamics of Contention:
Demobilization of the Antiwar Movement in the United States, 2007-2009." Mobilization
16:45-64.
Hofstadter, Richard. 1967. The Paranoid Style in American Politics and Other Essays.
New York: Vintage Books.’
Hopkins, Daniel. J. 2010. “Politicized places: Explaining where and when immigrants provoke
local Opposition”. American Political Science Review,104(01), 40-60.
Howard, Philip N, Aiden Duffy, Deen Freelon, Muzammil Hussain, Will Mari and Marwa
Mazaid. 2011. “Opening Closed Regimes: What was the role of social media during the
Arab Spring?” working paper
Huberty, Mark. 2013. “Voting with your tweet: Forecasting congressional elections with social
media data”. Midwest political science association conference, Chicago, Illinois.
Jackman, Robert W., and Karin Volpert. 1996. "Conditions Favouring Parties of the
Extreme Right in Western Europe." British Journal of Political Science 26(04):501-21.
Jamieson, Kathleen H. and Joseph N. Capella. 2008. Echo Chamber: Rush Limbaugh
and the Conservative Media Establishment. New York: Oxford University Press.
Java A, Song X, Finin T, Tseng B. 2007. “Why we twitter: understanding microblogging usage
and Communities”. In: Proceedings of the 9th WebKDD and 1st SNA-KDD 2007
workshop on Web mining and social network analysis. ACM, pp. 56–65.
172
Jenkins, J. Craig, and Charles Perrow. 1977. "Insurgency of the Powerless: Farm Worker
Movements (1946-1972)." American Sociological Review 42(2):249-68.
Jiobu, Robert and Timothy Curry. 2001. “Lack of Confidence in the Federal Government and the
Ownership of Firearms.” Social Science Quarterly 82: 77-88.
Jonsson, Patrik. 2010. “Why the Tea Party Convention is tea-tering on the edge”, Christian
Science Monitor, January 30. Retrieved May 12, 2012 from
http://www.csmonitor.com/USA/Politics/2010/0130/Why-the-Tea-Party-Convention-istea-tering-on-the-edge
Juris, Jeffrey S. 2012. “Reflections on #Occupy Everywhere: Social Media, public space, and
emerging logics of aggregation.” American Ethnologist 39 (2): 259-279
Karpowitz, Christopher F., J. Quin Monson, Kelly D. Patterson, and Jeremy C. Pope. 2011. "Tea
time in America? The impact of the Tea Party movement on the 2010 midterm
elections." PS Political Science and Politics 44, no. 2: 303.
Kazin, Michael. 1995. The populist persuasion: an American history. New York, NY:
BasicBooks.
Kennedy, Helen. 2010. “Tea Party Express leader Mark Williams Kicked out over
'Colored People' letter”, New York Daily News, July 18. Retrieved, May 16, 2012 from
http://www.nydailynews.com/news/politics/tea-party-express-leader-mark-williamskicked-colored-people-letter-article-1.438854
Khondker, Habibul. 2011. “Role of New Media in the Arab Spring” Globalizations 8 (5): 675-79
Lazer, David M., Ryan Kennedy, Gary King, and Alessandro Vespignani. 2014. "The parable of
google flu: traps in big data analysis." Science 343: 1203-1205
Lipset, Seymour Martin. 1960. Political man; the social bases of politics. Garden City,
173
N.Y.,: Doubleday.
Lipset, Seymour Martin, and Earl Raab. 1978. The politics of unreason : right-wing
extremism in America, 1790-1977. Chicago: University of Chicago Press.
Long, J. Scott. 1997. Regression Models for Categorical and Limited Dependent
Variables. London: Sage Publications.
Lubbers, Marcel, Mérove Gijsberts, and Peer Scheepers. 2002. "Extreme right-wing
voting in Western Europe." European Journal of Political Research 41(3):345-78.
Luker, Kristin. 1985. Abortion & the politics of motherhood. Berkeley and Los Angeles:
University of California Press
Madestam, Andreas, Daniel Shoag, Stan Veuger, and David Yanagizawa-Drott. 2013. "Do
Political Protests Matter? Evidence from the Tea Party Movement." The Quarterly
Journal of Economics 128, no. 4: 1633-1685.
McAdam, Doug. 1982. Political process and the development of Black insurgency, 19301970. Chicago: University of Chicago Press.
McAdam, Doug. 1986. "Recruitment to High-Risk Activism: The Case of Freedom
Summer." The American Journal of Sociology 92(1):64-90.
McCarthy, John D., and Mayer N. Zald. 1977. "Resource Mobilization and Social
Movements: A Partial Theory." The American Journal of Sociology 82(6):1212-41.
McVeigh, Rory. 1999. "Structural Incentives for Conservative Mobilization: Power
Devaluation and the Rise of the Ku Klux Klan, 1915-1925." Social Forces 77(4):146196.
McVeigh, Rory. 2001. "Power Devaluation, the Ku Klux Klan, and the Democratic
National Convention of 1924." Sociological Forum 16(1):1-30.
174
McVeigh, Rory. 2004. "Structured Ignorance and Organized Racism in the United
States." Social Forces 82(3):895-936.
McVeigh, Rory. 2009. The rise of the Ku Klux Klan : right-wing movements and
national politics. Minneapolis: University of Minnesota Press.
McVeigh, Rory, Daniel J. Myers, and David Sikkink. 2004. "Corn, Klansmen, and
Coolidge: Structure and Framing in Social Movements." Social Forces 83(2):653-90.
McVeigh, Rory, and David Sikkink. 2005. "Organized Racism and the Stranger."
Sociological Forum 20(4):497-522.
McVeigh, Rory. 2009. The rise of the Ku Klux Klan: right-wing movements and national
politics. Minneapolis: University of Minnesota Press.
Mitchell, Amy and Paul Hitlin. 2013. “Twitter reaction to events often at odds with overall
public opinion.” Recovered from http://www.pewresearch.org/2013/03/04/twitterreaction-to-events-often-at-odds-with-overall-public-opinion/
Morris, Jonathan. 2007. "Slanted Objectivity? Perceived Media Bias, Cable News
Exposure, and Political Attitudes." Social Science Quarterly 88:707-728.
Naaman M, Boase J, Lai CH. 2010. “Is it really about me?: message content in social awareness
Streams”. In: Proceedings of the 2010 ACM conference on Computer supported
cooperative work. New York, NY, USA: ACM, CSCW '10, pp. 189–192.
doi:10.1145/1718918.1718953.
O'Brien, Luke. 2010. “Judson Phillips Threw a Tea Party, and Trouble Came”, AOL News,
February 5. Retrieved May 16, 2012 from http://www.aolnews.com/2010/02/05/judsonphillips-threw-a-tea-party-and-trouble-showed-up/
O’Brien, Kerry, Walter Forrest, Dermot Lynott, and Michael Daly. 2013. "Racism, gun
175
ownership and gun control: Biased attitudes in US whites may influence policy
decisions." PlOS ONE 8(10): e77552.
O’Connor, Amy. 2010. Bringing the market back in: Philanthropic activism and conservative
reform, in Politics and Partnerships: The Role of Voluntary Associations in America’s
Political Past and Present, edited by E. S. Clemens and D. Guthrie.
Chicago: University of Chicago Press, 121–50.
Oesch, Daniel. 2008. "Explaining Workers' Support for Right-Wing Populist Parties in
Western Europe: Evidence from Austria, Belgium, France, Norway, and Switzerland."
International Political Science Review 29(3):349-73.
Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of Group.
Cambridge: Harvard University Press.
Orloff, Ann Shola. 1993. "Gender and the Social Rights of Citizenship: The Comparative
Analysis of Gender Relations and Welfare States." American Sociological Review
58(3):303-28.
Parker, Christoper and Matt Barreto. 2013. Change they can’t believe in: The Tea Party and
Reactionary Politics in America. Princeton, NJ: Princeton University Press
Perrin, Andrew J., Steven J. Tepper, Neal Caren, and Sally Morris. 2011. "Cultures of the tea
party." Contexts 10(2): 74-75.
Podsakoff, Philip M., Scott B. MacKenzie, Jeong-Yeon Lee, and Nathan P. Podsakoff. 2003.
"Common method biases in behavioral research: a critical review of the literature and
recommended remedies." Journal of applied psychology 88, no. 5: 879.
Potok, Mark. 2010. "Rage on the Right: The Year in Hate and Extremism." Southern
Poverty Law Center Intelligence Report (137). Retrieved May 16, 2012 from
176
http://www.splcenter.org/get-informed/intelligence-report/browse-allissues/2010/spring/rage-on-the-right
Presser, Stanley, and Linda Stinson. 1998. "Data collection mode and social desirability bias in
self-reported religious attendance." American Sociological Review : 137-145.
Quadagno, Jill. 1990. "Race, Class, and Gender in the U.S. Welfare State: Nixon's Failed
Family Assistance Plan." American Sociological Review 55(1):11-28.
Ratkiewicz, Jacob, Michael Conover, Mark Meiss, Bruno Goncalves, Snehal Patil,
Alessandro Flammini, and Filippo Menczer. 2011. "Truthy: mapping the spread of
astroturf in microblog streams." Pp. 249-252 in Proceedings of the 20th international
conference companion on World Wide Web. Hyderabad, India: ACM.
Rydgren, Jens. 2005. "Is extreme right-wing populism contagious? Explaining the
emergence of a new party family." European Journal of Political Research 4
4(3):413-37.
Rydgren, Jens. 2008. "Immigration sceptics, xenophobes or racists? Radical right-wing
voting in six West European countries." European Journal of Political Research 47:737765.
Scheufele, Dietram A., and David Tewksbury. 2007. "Framing, agenda setting, and priming: The
evolution of three media effects models." Journal of communication 57, no. 1: 9-20.
Skocpol, Theda. 1992. Protecting soldiers and mothers: the political origins of social
policy in the United States. Cambridge, Mass.: Belknap Press of Harvard University
Press.
Skocpol, Theda, and Vanessa Williamson. 2011. The Tea Party and the remaking of Republican
conservatism. Oxford University Press.
177
Skrentny, John David. 2002. The minority rights revolution. Cambridge, Mass.: Belknap
Press of Harvard University Press.
Smelser, Neil J. 1963. Theory of collective behavior. New York,: Free Press of Glencoe.
Snow, David A., Louis A. Zurcher, Jr., and Ekland-Olson Sheldon. 1980. "Social
Networks and Social Movements: A Microstructural Approach to Differential
Recruitment." American Sociological Review 45(5):787-801.
Staggenborg, Suzanne. 1998. "Social Movement Communities and Cycles of Protest:
The Emergence and Maintenance of a Local Women's Movement." Social Problems
45:180-204.
Steensland, Brian. 2006. “Cultural Categories and the American Welfare State: The Case of
Guaranteed Income Policy.” American Journal of Sociology,111(5), 1273-1326.
Stephens-Davidowitz, Seth. 2014. “The cost of racial animus to a black candidate: Evidence
using Google search data” Journal of Public Economics doi:
10.1016/j.jpubeco.2014.04.010
Stroud, Natalie J. 2007. "Media Use and Political Predispositions: Revisiting the
Concept of Selective Exposure." Political Behavior 30:341-366.
Swank, Duane, and Hans-Georg Betz. 2003. "Globalization, the welfare state and right-wing
populism in Western Europe." Socio-Economic Review 1(2): 215-245.
Swearingen, C. Douglas and Joseph T. Ripberger. 2014. “Google Insights and U.S. Senate
Elections: Does Search Traffic Provide a Valid Measure of Public Attention to Political
Candidates?” Social Science Quarterly DOI: 10.1111/ssqu.12075
Tarrow, Sidney G. 1993. "Cycles of Collective Action: Between Moments of Madness
and the Repertoire of Contention." Social Science History 17:281-307.
178
Tarrow, Sidney G. 1998. Power in movement : social movements and contentious
politics. Cambridge England; New York: Cambridge University Press.
Taylor, Marylee C. 1998. "How white attitudes vary with the racial composition of local
populations: numbers count." American Sociological Review 63: 512-535.
Tea Party Patriots Mission Statement and Core Values, n.d. Retrieved from:
http://www.teapartypatriots.org/mission.aspx.
Tucker, Katheryn Hayes. 2011. “Tea Party Patriots in Intellectual Property Battle with ExMember”, Fulton County Daily Report, December 9. Retrieved May 16, 2012 from
http://www.law.com/jsp/cc/PubArticleCC.jsp?id=1202534872589&Tea_Party_Patriots_i
n_Intellectual_Property_Battle_with_ExMember
Tumasjan, Andranik, Timm Oliver Sprenger, Philipp G. Sandner, and Isabell M. Welpe. 2010.
"Predicting Elections with Twitter: What 140 Characters Reveal about Political
Sentiment." ICWSM 10: 178-185.
Van Dyke, Nella and Sarah Soule. 2002. "Structural social change and the mobilizing effect of
threat: Explaining levels of patriot and militia organizing in the United States." Social
Problems 49:497-520
Vogel, Kenneth 2009. “Tea parties emerge as revenue stream.” Politico, November 27.
Retrieved May 16, 2012 from
http://news.yahoo.com/s/politico/20091127/pl_politico/29943/print
Vogel, Kenneth. 2010. “Tea Party Nation leaders lash out over convention”, The Virginian-Pilot,
February 1. Retrieved May 12, 2012 from http://hamptonroads.com/2010/02/tea-partynation-leaders-lash-out-over-convention
Vosen, Simeon and Torsten Schmidt. 2011.“Forecasting private consumption: survey-based
179
indicators vs. Google trends.” Journal of Forecasting 6(30):565-578
Walker, Edward T. 2009 "Privatizing participation: Civic change and the organizational
dynamics of grassroots lobbying firms." American Sociological Review 74.1: 83-105.
Williamson, Vanessa, Theda Skocpol and John Coggin. 2011. “The Tea Party and the Remaking
of Republican Conservatism.” Perspectives on Politics 9:25-43
Zernike, Kate. 2010. Boiling Mad : Inside Tea Party America. New York: Times
Books/Henry Holt and Co
180
June 2014
Joseph DiGrazia
Department of Sociology
1020 East Kirkwood Ave, Room 744
Bloomington, IN 47405
[email protected]
Position
2014-
Neukom Fellow, Neukom Institute and Department of Sociology,
Dartmouth College
Education
2014
Ph.D. Sociology, Indiana University, Bloomington
Minor: Quantitative Research Methods
Dissertation: The Tea Party Movement: Right-Wing Mobilization in the
Age of Obama
Committee: Fabio Rojas (chair), Art Alderson, Patricia McManus, Rob
Robinson, Tim Bartley (Ohio State University)
2009
Qualifying Examination: Political Sociology
2008
M.A., Sociology, Indiana University, Bloomington
2005
B.A. Anthropology (honors), University of Notre Dame
Areas of Interest: Political Sociology, Social Movements, Conflict, Policy, Quantitative
Research Methods, Computational Social Science, Race
Peer Reviewed Publications
McKelvey, Karissa, Joseph DiGrazia and Fabio Rojas. Forthcoming. 2014. “Twitter Publics:
How online political communities signal electoral outcomes in the 2010 U.S. House Elections.”
Information, Communication and Society 17: 436-450
DiGrazia, Joseph. 2014. “Individual Protest Participation in the United States: Conventional and
Unconventional Activism.” Social Science Quarterly 95:111-131
DiGrazia, Joseph, Karissa McKelvey, Johan Bollen and Fabio Rojas. 2013. “More Tweets, More
Votes: Social Media as a Quantitative Indicator of Political Behavior.” PLOS ONE 8(11):
e79449. doi:10.1371/journal.pone.0079449
Manuscripts Being Revised for Resubmission
DiGrazia, Joseph. “Explaining Tea Party Activism: Cultural, Political and Economic Threat in
the Policy Stream.” Revise and Resubmit at Mobilization
Manuscripts in Progress
DiGrazia, Joseph. “The Shifting Basis of Tea Party Support: Evidence from polling data.”
DiGrazia, Joseph. “The impact of Tea Party Mobilization on Policy.”
DiGrazia, Joseph. “Anti-Immigrant Sentiment and the Tea Party Movement: Racist web searches
as a predictor of Tea Party mobilization.”
DiGrazia, Joseph and Shiri Noy. “An investigation of the utility of novel data sources in
constructing economic measures.”
Selected Media Coverage
My work on Twitter and election outcomes has been covered in The Washington Post, The
Huffington Post, The Wall Street Journal, The Atlantic, Popular Science, and National Public
Radio.
Honors and Awards
2014
Lindesmith-Mullins Fellowship For Excellence in Research,
Department of Sociology, Indiana University
2012
Service-Learning Graduate Fellowship, Center for Innovative
Teaching and Learning, Indiana University
2012
Lindesmith Travel Fellowship, Indiana University
2008
Schuessler Scholarship for Study at ICPSR
2008
Summer Research Fellowship, Department of Sociology, Indiana
University
2006-present
Graduate Tuition Scholarship, Indiana University
Presented Papers/Invited Talks
2013
“More Tweets, More Votes: Social Media as a Quantitative Indicator of
Political Behavior.” American Sociological Association, New York, NY
2012
“Explaining Tea Party Activism: Cultural and Economic Threat.” American
Sociological Association, Denver, CO
2012
“Dissent at a Distance.” Invited discussant at Media Arts & Sciences Speaker
Series, Department of Telecommunications, Indiana University
2010
DiGrazia, Joseph. “Individual Protest Participation in the US: High-Risk/Cost
and Low-Risk/Cost Activism. American Sociological Association, Atlanta,
GA
2008
DiGrazia, Joseph. “Exploring the Relationship Between Protest and
Government Confidence.” American Sociological Association, Boston, MA
2008
DiGrazia, Joseph. “Confidence and Protest Participation” North Central
Sociological Association, Cincinnati, OH
Teaching Areas
Statistics and Quantitative Methods, Introduction to Sociology, Theory, Political Sociology,
Social Movements
Teaching Experience
LAMP 316: Analytical Problem Solving: an undergraduate service learning course in
statistics, taught in the Liberal Arts and Management Program (Fall 2011, Spring 2012,
Fall 2012, Spring 2013)
Sociology 554: Statistical Techniques for Sociology I (Lab session): a graduate level
course on regression and regression diagnostics (Spring 2011)
Sociology 566: Sociological Research Practicum (Associate Instructor): a graduate-level
course designed to help first year graduate students begin developing master’s thesis
projects.
Sociology 371: Statistics for Sociology: an undergraduate introductory statistics course
(Spring 2010, Summer 2010, Fall 2010, Summer 2011, Summer 2012, Summer 2013).
Sociology 100: Introduction to Sociology: an undergraduate introductory sociology
course (Fall 2008).
Indiana Intensive Didactic Seminar (IIDS)
Statistical Computing in Stata: a one-day graduate-level seminar on issues in statistical
computing using Stata (Spring 2010, Spring 2011, Spring 2012).
Graduate Teaching Assistant
Spring 2008
Sociology 100: Introduction to Sociology
Dept of Sociology, Indiana University, Professor Rob Robinson
Fall 2007
Sociology 100: Introduction to Sociology
Dept of Sociology, Indiana University, Professor Paulette Lloyd
Spring 2007
Sociology 338: Gender Roles
Dept of Sociology, Indiana University, Professor Brian Powell
Fall 2006
Sociology 339: Media and Society
Dept of Sociology, Indiana University, Professor Christine Von Der
Haar
Research Experience
2009
Associate Instructor, Sociological Research Practicum . Clem
Brooks (Principal Investigator). Duties: Contributed to design of
survey instrument for a national telephone survey and supervised
interviewers.
2008
Research Assistant, Tim Bartley (Principal Investigator). Primary
Duties: Collecting and Synthesizing information on Indonesian
Labor and Forest Certification movements
2007
Research Assistant “Survey of the American Antiwar Movement,”
Fabio Rojas (Principal Investigator). Primary Duties: Collecting
survey data from protesters in the field.
2007
Interviewer “Sociological Research Practicum”, Clem Brooks
(Principal Investigator). Primary Duties: Interviewing respondents
over the telephone.
2005-2006
Research Assistant “Oppositional Consciousness Research
Project” Erika Summers-Effler (Principal Investigator). Primary
Duties: Conducting in-depth interviews, coding data.
Reviewer For:
Mobilization: An International Journal
PLOS ONE
Social Science Computer Review
Professional Service
2010-2011
Graduate Student Association, Research Infrastructure
Representative
2010
Mentor Award Committee, Department of Sociology, Indiana
University
Spring 2010
Graduate Student Association, Institute for Social Research
Representative
November 2009
Judge, Indiana University Undergraduate Research
Conference
2009-2010
Mentor, Graduate Student Mentoring Program
2009
Organizer for NCSA Political Sociology Session
2008
Organizer for NCSA Political Sociology Session
2007-2008
Mentor, Undergraduate Honors Thesis Mentoring Program, Department of
Sociology, Indiana University
Professional Associations
American Sociological Association (2008 to present)
American Political Science Association (2013 to present)
North Central Sociological Association (2008 to 2009)
Midwest Sociological Association (2008, 2012)
Software and Computing
Statistical Software Competence: HLM, Mplus, R, SPSS, Stata
Programming languages: Python, R, Ruby
Other Scientific Software: LaTeX
References:
Fabio Rojas
Associate Professor
Department of Sociology
Indiana University
744 Ballantine Hall
1020 E. Kirkwood Ave.
Bloomington, IN 47405
Phone: 812-856-1419
Email: [email protected]
Patricia McManus
Associate Professor
Department of Sociology
Indiana University
744 Ballantine Hall
1020 E. Kirkwood Ave.
Bloomington, IN 47405
Phone: 812-855- 8970
Email: [email protected]
Robert Robinson
Chancellor's Professor
Department of Sociology
Indiana University
744 Ballantine Hall
1020 E. Kirkwood Ave.
Bloomington, IN 47405
Phone: 812-855-3987
Email: [email protected]
Johan Bollen
Associate Professor
School of Informatics and Computing
Informatics East 305
919 E. 10th Street
Indiana University
Bloomington, IN 47401
Phone: 812-856-1833
Email: [email protected]