=Paper=
{{Paper
|id=Vol-1710/paper16
|storemode=property
|title=Organizations, State Interactions, and Field Stability: A Network Analysis of the Tea Party and Occupy Wall Street Movements
|pdfUrl=https://ceur-ws.org/Vol-1710/paper16.pdf
|volume=Vol-1710
|authors=Benjamin Lind,B. Remy Cross,Georgy Mkrtchyan
|dblpUrl=https://dblp.org/rec/conf/aist/LindCM16
}}
==Organizations, State Interactions, and Field Stability: A Network Analysis of the Tea Party and Occupy Wall Street Movements==
https://ceur-ws.org/Vol-1710/paper16.pdf
State Embeddedness and Field Stability: The Tea Party and Occupy Wall Street's
Interorganizational Networks
Benjamin E. Lind1, B. Remy Cross 2, and Georgy Mkrtchyan1
1
School of Sociology and the Applied Network Research Laboratory, National Research UniversityHigher School of
Economics, Moscow, Russian Federation
{blind,gnmkrtchyan}@hse.ru
2
Department of Sociology and Criminology, Webster University, St. Louis, Missouri, United State
briancross20@webster.edu
Abstract. Despite widespread scholarly attention on social movement media coverage, the full range of organizations
appearing in the media with movements remains underexamined. To analyze this organizational complexity, we draw upon
field theory, relating it to concepts from social network analysis. We hypothesize that interactions with stateembedded
actors increase field stability, yet this effect decreases with the field's level of abstraction. We contrast a pair of US
movements, the Tea Party and Occupy Wall Street, according to their interactions with stateembedded actors. Through the
New York Times's API, we collect and produce article coappearance networks that approximate the population of Times’s
indexed organizations. We parse fields from these networks using community detection algorithms. Our analyses test
stability using triadic closure propensities and average coreness. Our findings show that subgroups affiliated with the Tea
Party had greater stability than those of Occupy Wall Street.
Keywords: Social networks · News media · Social movements · Organizations
1 Introduction
Social movements research using data from news media sources has recently blossomed. This research has expanded upon
diverse topics that include movement media coverage, tactics, policy and market outcomes, and targets [e.g., 3, 28, 53].
Within this large research body, two subjects remain relatively unexamined: the ubiquitous presence of nonmovement
organizations and their social networks, which may or may not include movement actors.
Scholarship has increasingly emphasized interactions between social movements and nonstate organizations. Calls
for such emphases span decades [11, 48] and have been reinvigorated by integrating social movements within field theory
and multiinstitutional politics [4, 14, 19, 20, 47]. Empirical progress demonstrating the interaction between movement and
nonmovement organizations has made great strides in analyzing relationships between movements and market actors [29]
among other nonmovement organizations [16]. Works demonstrating such interaction, typically focus upon nonmovement
actors as entities vulnerable to movement action (e.g., targets), supporters or adversaries, or instruments that voice
movement interests. While these studies move beyond the “movementcentric” criticisms of movement theory [19, 20], their
research designs largely place relational emphases on a “twoparty struggle” [48:197], ignoring a vast number of
organizations within the broader field of action.
Social network theory and analysis provides one way to capture such organizational complexity [34]. Scholarship
on interorganizational network has unquestionably advanced social movements research [e.g., 13, 14, 32, 53]. Of these
studies, two works that used newspaper data to construct social networks [5, 53] restricted their analyses to movement
organizations and protest events. Other remaining works analyze interorganizational networks that are smalltomoderate
sized in terms of the number of organizations and, exempting one study [41], focus on university or cityspecific research
universes, rendering generalizations to newspaper coverage on a larger scale difficult.
We collect and analyze two networks constructed from newspaper data that approximate all organizations
recognized by the New York Times (NYT) during the analyzed time periods. Each network spans a coverage period
associated with two movements: the Tea Party (TP) and Occupy Wall Street (OWS). To introduce our guiding theoretical
framework, we briefly review the literature on fields. Given implications in the literature, we suggest that interactions with
stateembedded actors stabilizes fields. Due to TP’s formal inclusion into US national politics and OWS’s ultimate
repression, we suspect that fields associated with TP will offer greater stability. Following new institutionalist studies, we
identify fields based upon dense network ties and use structural cohesion to characterize the stability within the fields [44].
We analyze field stability alongside “movementcentric” subgroups generated from network neighborhoods. We conclude
our study by encouraging future social movement studies to engage in multiinstitutional research and expand upon similar
network datasets.
2 Organizations, Fields, and Social Movements
Scholarship on organizations have benefited greatly due to the popularization of the organizational fields concept, defined as
the “sets of organizations that, in the aggregate, constitute a recognized area of institutional life” [15:1489]. Here, we
understand institutions as sets of rules that develop through social interactions [18:106].
The concepts of fields and institutions have proven resonant among social movement scholars [47, 4, 14, 19, 20].
Synthesizing two disparate approaches, one study characterizes a field as having regular (inter)action within itself,
agreement among participants regarding the “rules of the game,” and potential vulnerability to field external forces [47:124].
An additional study encouraged analysts to consider the many nonstate institutions with whom movements interact [4], a
point strongly emphasized by a later study which conceptualized fields “as sets of actors characterized by high relational
density and actors’ reciprocal recognition” [14:265]. Recent work on “strategic action fields,” defined as “a mesolevel social
order where actors...interact with knowledge of one another under a set of common understandings about the purposes of
the field, the relationships in the field..., and the field’s rules” [19:3], has operationalize field membership as “those groups
who routinely take each other into account in their actions” [20:1678]. Though differing in some ways, the literature
converges upon characterizing a field based on frequent interactions between recognizable actors who moreorless
understand the field’s “rules.”
In addition to this characterization, scholars have describe fields and organizational networks as a nested
phenomenon [4:82, 19, 20]. “Strategic action fields,” for instance, are generally couched within other strategic action fields,
like the nested layers of Russian matryoshka dolls [19, 20]. Though vocabulary differs between studies, the central premises
hold that smaller fields are embedded within larger fields and that smaller fields are more specialized, yielding more similarities
among their member organizations than their larger, enveloping fields. This process of field aggregation will therefore
accompany heterogeneity in both its interaction patterns and understood rules.
Though movements may engage within any of a vast array of fields, analysts must consider structural variation
between fields. Recent scholarship has presented three basic field phases: emergence, crisis, and stability [19, 20]. Of these
three phases, we are most interested in contrasting emergent and stable fields. Actors in emergent fields possess limited
agreement regarding field rules, maintain tenuous alliances with actors from other fields, are unsure regarding challenger
outcomes, and are vulnerability to state action. On the flipside, stable fields follow wellestablished, known rules of an
“ongoing game” in which incumbents, challengers, and their political coalitions jockey for improved positions through
iterative moves and countermoves. A field’s stability hypothetically arises from many ties to other fields, invulnerability to
turbulences in its resource dependencies, and its relationship to the state, an institution with the capability to reshape field
rules [19, 20].
Fields comprised of organizations with frequent ties or longstanding relationships to actors embedded in state
affairs likely enjoy greater stability than those without. We reach this hypothesis from three main points. First, we should
consider that stateembedded actors may or may not even interact with challengerseven if the challenger orients its action
to the state. In the case of the United States, such interactions typically coincide with national elections, times when
challengers compete to influence political party platforms [2:292]. Second, given the descriptions of field phases [19, 20], we
suggest that highcapacity nation states generally constitute stable fields, as they meet the characteristics of emergent fields
only following revolutions. Third, among professions, one study [15] points to the role of the state in driving organizational
isomorphismthe process by which organizations and their fields become similar in form. Two of this study’s fieldlevel
hypotheses posit that a field’s resource dependence and transaction with state agents yield higher levels of isomorphism
across all organizations within the field. To the extent that stateembedded actors like the Republican and Democratic parties
include challengers, we expect that challenger inclusion corresponds with frequent ties and a high degree of isomorphism for
both challengers and incumbents. If we characterize state fields as stable, fields that depend upon and frequently engage
with the state should therefore adopt stable forms more than other fields. Lastly, following fields’ nested characterizations, as
a field expands to include relatively fewer interactions with and dependence upon stateembedded actors, its stability should
diminish.
3 Case Overviews
For this paper, we contrast two cases, the Tea Party and Occupy Wall Street. We focus on these two movements due to their
large size, national media coverage, and historical proximity. Though other studies have compared these two movements,
such as their approaches to populism [54] among others, ours highlights their relationships to other movement and
nonmovement organizations.
3.1 Tea Party
The Tea Party was initially hailed as a spontaneous grassroots movement that emerged in early 2009 to oppose a variety of
issues including the election of Barack Obama, proposed health care reform, and increased taxation in early 2009 [55]. As TP
grew in popularity it expanded the initial focus on resisting increased government services, encompassing a wide array of
traditionally conservative causes including taxation, smaller government, along with wider economic and social issues such
as immigration, welfare, and affirmative action.
The nature of TP protests is contentious, with their members making claims of being a wholly grassroots movement
[40] and opponents claiming that the movement is mostly ‘astroturf,’ a fake grassroots movement completely backed and
orchestrated by monied interests [55]. The narrative advanced by TP organizations and most affiliated groups is that the
movement is “100% grassroots, 100% of the time” [50] a stance that allows the movement to frame itself as a scrappy
outsider unconnected to current institutional insiders.
Those arguing that the Tea Party grew out of, and enjoyed the support of, existing political organizations and
movements point to the significant resources from established political interests including the Republican Party and think
tanks like Americans For Prosperity and FreedomWorks, all of which had activists and resources invested in the movement
[55]. This support is credited as helping to elect members of the “Tea Party Caucus” in both the House and Senate in the
2010 elections. Additionally, TP groups often are financially supported by industry and conservative think tanks like
Americans for Prosperity and the Heartland Institute. But TP groups are often at odds with mainstream Republican policies
and are more likely to support libertarian ideals and causes when these causes are in conflict with a broader conservative
agenda.
This second account seems more supported by the gradual movement of the Tea Party towards alliances with
institutionally powerful organizations and political parties. Despite their claims of local grassroots control, TP was able to
quickly mobilize or recruit several members who successfully ran for public office at both the state and federal level in 2010
with an even greater number running and failing to win public office. This would seem to indicate that TP was able to make
linkages with institutionally powerful organizations within its field, and not just constraint itself to decentralized local
chapters with few ties at the national level.
Our data on TP’s media presence covers the period from February 27, 2009 which was the date of the first national
TP protests, to November 2, 2010 the date of the first general elections in which the movement played a significant role.
3.2 Occupy Wall Street
Occupy Wall Street represents the most recent incarnation of the long lived tradition of movements focused on social justice
and the alleviation of income and other inequalities. Initially formed in September of 2011 under the auspices of the Canadian
anticonsumerist group Adbusters, the OWS movement held its first major protest event on September 17, 2011 in New York
City’s Zuccotti Park. However, its roots trace inspiration from the Spanish economic protests of the prior May, which in turn
were inspired by the events of the prodemocracy Arab Spring [26].
During this period of time OWS spread to a plethora of other cities and gained considerable attention of the
national news media. The period witnessed the emergence of other social movement groups who addressed issues pertinent
to OWS’s interests, including hacktivist groups like Anonymous [8], as well as other social justice and traditional leftwing
groups. During the early days of the movement these diverse interests vied to set the agenda of the overall movement and
attempted to link the overall movement goal of inequality with a myriad of local and national issues [46]. Occupy Wall Street
was also quick to rebuff overtures from national political figures during this time, instead choosing to allow local occupation
groups to retain a high degree of autonomy [21].
Our data encompass the first organized protest on September 17, 2011 until March 17, 2012. This end date marks the
day six months after the start of the movement and corresponds with Occupiers’ attempt to reoccupy Zuccotti Park, in
which the group faced mass arrests and violence from police. This attempted reoccupation followed a period of general
movement decline after police vigilantly cleared Occupy encampments across the country.
4 Methods of Analysis
4.1 Data
To take census of the fields present in the media, our study relies upon a network unique dataset collected from the New
York Times Article Search API online [38].1 The NYT Article Search API enables researchers and developers the ability to
search all articles published by the New York Times between 1981 and the present. Beyond conventional search capabilities
to identify articles by keyword strings contained in the text, the NYT Article Search API [38] also enables “facet” searches.
Here, a facet refers to a type of tag used to index New York Times articles. The API’s facets provide information on a
descriptive subject (des_facet), relevant geographical information (geo_facet), the standardized names of organizations
mentioned in the article (org_facet), notable persons mentioned (per_facet), among many other characteristics relevant to
the article’s classification. We focus almost exclusively upon the organizations tagged in each article using R version 2.14.1
script [45] with the packages RCurl [31] and rjson [9].
Generally, the New York Times indexers standardized facet names. To identify a concept’s standardized name and
relevant facets within the Times one must use the TimesTags API [39].We searched the TimesTags API for the terms “tea
party” and “occupy” to find articles mentioning the TP and Occupy Wall Street. The API respectively returned “Tea Party
Movement (Des)” and “Occupy Wall Street (Org),” indicating that we should only search for “Tea Party Movement” under
the descriptive subject facet and “Occupy Wall Street” under the organizational facet. We retrieved all articles on TP
between February 27, 2009 and November 2, 2010 and all articles on OWS between September 17, 2011 and March 17, 2012.
The New York Times identified TP in 207 articles and tagged OWS in 294, after removing duplicates. Despite early movement
coverage, the newspaper only began tagging TP on January 21, 2010. Taking TP’s late tag implementation into consideration,
it’s little surprise that OWS received more coverage due to its point of origin and focus relative to the Times.
After we collected the articles tagging the Tea Party or Occupy Wall Street, we searched for a near population of
articles with organizational tags through a crawling technique. The technique followed a five step process. First, we
identified additional organizations from the organizational facets of the initial articles on TP and OWS. We did so separately,
creating two lists: one of organizations tagged in TP articles and one of organizations tagged in OWS articles. We refer to
these organizations as the sample’s “first wave.” Second, we search for all articles tagging these additional organizations. If
we found additional organizations from the TP articles, we restricted the results to articles running between February 27, 2009
and November 2, 2010; for organizations discovered from OWS articles, we only searched for additional articles between
September 17, 2011 and March 17, 2012. Third, we delete duplicate articles identified by the previous searches. Fourth, we
identified additional organizations uncovered from the organizational facets of the first wave. These organizations constitute
the sample’s second wave. Fifth, we repeated the second through fourth steps until the fourth step could no longer identify
any additional organizations. The process took eight waves for TP and seven waves for OWS. While this procedure cannot
capture every organization ever tagged in the New York Times during these periods, it does capture all organizations tagged
in the giant component of the New York Times’s organizational network. Of great value to social movement research, the
method generates a grounded, systematic list of organizations with potentially distant relations to the movements of interest.
1
We abide by the terms of use put forth by the Article Search API [38]. In doing so we must acknowledge that the New York
Times neither promotes nor endorses our research.
While the technique to use “facets” through a newspaper’s API is new to social movements research, it offers a
technological update to established methods in the movements literature which used article indexing [e.g., 33:23650, 33:267].
This method faces limitations which include practices of inclusiveness, thoroughness, and consistency exercised by
newspaper indexers [17:68]. Rather than focusing upon protest coverage, the subject of criticism [17:68], our study addresses
general patterns of organizational affiliations through mutual media coverage.
Following the data collection stage, we processed the data into two social networks. This stage included refining
the list of organizations provided by the New York Times and converting organizational comentions within an article to
edges (i.e., “ties” or “relationships”) in a social network. We split the data into two social networks: one created during the
the Tea Party Movement’s time span and the other collected during Occupy Wall Street’s window. The Article Search API
[38] produced a total of 4,852 organizations from the TP crawl and 2,923 organizations from the OWS crawl. Together, 6,330
organizations appeared in either crawl, leaving 1,445 organizations appearing in both. However, these counts include
duplicate entries which the authors consolidated. Some “organizations” failed to meet our very general criteria for inclusion,
though. These criteria include one or more living member dedicated to the maintenance of the organization and its purposeful
activities, as well as a modest degree of independence or distinction in terms of personnel, purpose, or clientele from other
organizations. Further, an organization must have the potential to exert agency. This criteria includes both subsidiaries and
parent organizations. It removes products, projects, services, and websites, should their existence depend exclusively upon
representing another organization. We err on the side of overinclusion, as closely related organizations should exist within
the same field. We are then left with 4,504 organizations in the TP data and 2,825 in the OWS data, for a total of 5,812 unique
organizations.
After determining organizational inclusion, we then created our networks, such that vertices represent organizations
and edges portray each article that includes facets referencing two or more organizations. Our data includes the possibility
for multiple edges between a pair of nodes, as two organizations are often cotagged in more than one article. We do exclude
“loops,” or relationships a node has with itself. We stored our network data and conducted all network analyses in this paper
using igraph (version 0.6) for R [10].
4.2 Community Detection
Community detection identifies groups within a network that contain many ties among the group’s members and few ties
between groups [36]. Scholars refer to this property of dense interaction within communities and sparse ties between as
“modularity” [35]. Though the techniques differ, similar motivations emphasizing dense ties within groups have motivated
other social movements analyses on group identification [5, 13, 14] and the study of fields [44]. As with field scholars,
community detection values dense interactions among actors within groups. While community detection does not assuredly
produce fields in which actors understand “the rules of the game,” the underlying premise behind this research is that
“tightknit” communities include members united by common attributes or themes [12]. Considering the definition of
institutions [18:106], mutual rule understandings should emerge from groups marked by frequent interactions. This approach
follows research that explicitly eschews a priori assumptions regarding institutional power, connections, and logics along
with field boundary definitions ascribed by actor attributes [4:82 and 14]. For this article, we refer to the communities
identified as “fields.”
Though field theorists add analytic complexity by conceptualizing their subject as a nested phenomenon,
community detection researchers have developed techniques for this very problem [6]. “The Louvain method” detects
communities by optimizing modularity in a hierarchical manner, looping through levels of aggregation. In the first pass, the
method identifies many small, dense communities. Following this pass, the method then treats these communities as
aggregated nodes within the network and identifies communities among these communities, or “metacommunities” [6:5].
The method repeats, treating these metacommunities as aggregated nodes, and so on, reducing the number of communities
until the algorithm reaches optimal modularity. Each pass represents a distinct level of communities within communities
differing from random association. The first level includes the most specific communities that feature the highest degree of
association and this degree of association wanes as the levels increase.
4.3 Subgroups
We focus upon two subgroup types. Communities detected by the Louvain method serve as the first type. We apply this
method to both TP and OWS datasets, allowing the method to interpret multiple edges (i.e., organizations cofaceted in
multiple articles). Due to the method’s focus on dense interaction patterns and nestedness, we liken Louvain communities to
fields and use the terms synonymously throughout our analyses. We analyze the communities at each relevant level of the
algorithm’s aggregation, whereby higher levels indicate greater abstraction and relatively less interaction within groups than
lower levels.
“Neighborhoods” serve as the second subgroup type. Neighborhood subgroups contain all actors tied to one focal
actor along with all ties between them. Neighborhoods vary in “order” so as to include indirect ties to the focal actor. A first
order neighborhood includes only those organizations with direct ties to TP or OWS. A second order neighborhood
includes all organizations who belong to order one, as well as all other organizations with whom they share direct ties. Lastly,
a third order neighborhood includes all organizations with direct ties to those in a second order neighborhood. We analyze
the first through third order neighborhoods. Like levels within Louvain communities, as the order of a neighborhood
increases, so too does the abstraction from its points of origin. Because we use two movements as the focal actors in these
networks, we refer to these analyses as “movementcentric.”
We operationalize subgroups this way for two reasons. First, we find value in comparing “movementcentric” and
“fieldcentric” analyses. Louvain communities and neighborhoods offer a different perspectives to analyze distance from a
movement’s core. Louvain communities express this distance in levels marked by a decrease in tie density between actors.
Neighborhoods express this distance in terms of path length from a focal actor. As the path length between actors increases,
similarities should diminish. Second, these comparisons also offer verification against idiosyncratic properties of either
method.
4.4 Stability
Among the concepts in social network analysis used to assess stability, cohesion ranks top. In principle, cohesion enables a
subgroup to readily resist “disruptive forces” [25], generate social homogeneity [22], and enforce norms through social
pressure [23:4189]. While cohesion definitions vary [23:410], practically all analysts agree that cliques constitute a cohesive
subgroup. A clique refers to a subgroup of size three or greater in which every member has a direct tie to every other member
[51:254]. Triangles constitute the smallest clique and their closure has received considerable attention in numerous classic
works [24; 27]. Analyzing closed triads, rather than dyads, offers at least three advantages relevant for this study [30:185].
First, the concept of a majority exists in groups of three or more, suppressing individual interests in favor of group interests.
Second, no single member in closed triangles wields the power to dissolve the group should his or her demands go unmet.
Third, groups of three address and resolve conflicts more readily than dyads due to thirdparty moderating effects. For these
reasons, we use the proportion of triangles, the global clustering coefficient [43:2], on our simplified networks to measure of
group stability. We apply this measure to each community.
Though most network analysts agree that cliques constitute a cohesive subgroup, most also consider the definition
too “stingy” [23, 51] as it excludes nearcliques and most large groups. The concept of kcores emerged as one popular
responses to cliquebased limitations by identifying dense, cohesive seedbeds within a network. k cores measure cohesion
by counting the minimum number of ties, k , each member of a subgroup has to other members within the subgroup [49,
51:2667]. Likewise, “coreness” refers to a nodal property represented by c, where c captures if the node belongs to a ccore,
but not a (c+1)core [1]. Coreness represents the highest k core a node belongs, where “highest” is accounted for by k .
Higher values of k and coreness indicate greater cohesion. We simplify our networks, allowing up to just one tie between
actors, and contrast TP and OWS’s subgroups according to their average coreness scores. High average values of coreness
indicate stability within the subgroup.
5 Analysis
5.1 Field Descriptions
We begin our analyses with an overview of the Louvain communities (“fields”) associated with TP and OWS. In Table 1 we
describe the subgroups associated with the two movements at each level. We provide counts for the number of
organizations associated with each cohesive subgroup, the proportion of observed ties (i.e., density) within each subgroup,
and name the five organizations added to the subgroup who received the most news coverage. Due to the number of
organizations included in each field, we mention only a few widely covered ones in the table and name others in the text.
Table 1. Field Overview
Case Level n Density Examples
TP 1 107 0.045 Democratic Party, Republican Party, Tea Party, City Council, Park51
TP 2 385 0.013 Senate, House of Representatives, Supreme Court, EPA, Harvard
TP 3 390 0.013 NSF, NIST, American Assoc of University Women, Kennedy School of
Government
TP 4 390 0.013
OWS 1 60 0.046 Occupy Wall Street, Police Dept, NYT, FBI, City Council
OWS 2 145 0.020 Lincoln Center for the Performing Arts, Fire Depart, Anonymous, Juilliard
School, Academy of Motion Picture Arts and Sciences
OWS 3 204 0.014 Education Depart, Olympus Corp, Parks and Rec Dept, Race to the Top
Fund, Board of Regents
Like the concept of fields, the number of organizations within a Louvain community grows as the level of
abstraction increases. Level 1 refers to the most specific group whose commonalities should be strongest according to the
Louvain method’s algorithm. Level 2 includes all organizations from Level 1, along with other related organizations whose
Level 1 group differed from our movements of interest. Likewise Level 3 includes all organizations from Level 2 and also
organizations from Level 2 groups that TP and OWS did not belong to. For OWS data, the Louvain method reached optimal
modularity at Level 3, so a Level 4 does not exist in its network. The technique did suggest a Level 4 pass for TP’s network
data, yet this pass merged only two groups and did not affect the Louvain communities associated with the Tea Party: TP’s
Level 4 community contains the same organizations as its Level 3 community. For this reason, we exclusively discuss Levels
1 through 3.
The Tea Party belongs to stateembedded fields at each field level. As much of its faceted coverage pertained to the
2010 primaries and general election, naturally its Level 1 subgroup heavily represents political parties. In addition to the
Republican and Democratic parties, this field also includes minor political parties and related organizations, such as the
Democratic and Republican national committees. Notably, this field also includes some of the largest and most
widelycovered US social movement organizations [3], such as the NAACP, the AARP, and the Sierra Club. It also involves
large labor organizations, health and medical groups, and conservative think tanks. Broadly described, this field
encompasses parties and movements successfully included in the political process. This field of national political contenders
lies within two other larger fields with less welldefined boundaries. The level of abstraction from Level 1 to Level 2 marks a
growth in over three and a half times as many organizations. In comparison, we observe a difference from TP’s Level 2 to
Level 3 field: the Level 2 field merges with just five organizations. Given the dramatic growth from Level 1 to Level 2, the
diversity observed with TP’s Level 2 field exceeds our ability to characterize its organizations. We would like to point out a
few organizations whose affiliations with TP’s most specific community, at Level 1, prove natural. One example of these
include governmental organizations, such as two of the three branches of the federal government, federal agencies, and
cabinet departments. These are, after all, the organizations that political parties seek to attain and politicallyoriented national
movements desire to influence. While many movement organizations do exist within TP’s Level 2 community, their presence
in this field is accompanied by lobbyists, elite private universities, energy firms, pharmaceutical companies, nonprofit
organizations, and other organizational types, many vying for political influence. Although these organizations frequently do
influence the polity, their position exists one step beyond national political parties and electoral politics.
In contrast to the Tea Party, OWS’s fields have relatively little to do with nationallyoriented politics. Their most
specific community, Level 1, includes economic authorities, organizations related to parks and their occupation, churches,
news agencies, California universities, online communities, and law enforcement agencies. The field did include some
movement organizations along with apolitical and economic targets, support organizations and venues, news media groups,
and authorities. Though some nationallevel organizations do exist within this field, by and large the field operates within a
much smaller locus of conflict. As with TP, OWS’s field diversity increases dramatically between the first and second levels.
Its second level includes 30 fields identified from the first pass and many of them quite small, precluding the level from
systematic characterization. In addition to OWS’s Level 1 field, its Level 2 field included seven other Level 1 fields whose
sizes exceed three organizations. These groupings included a number of organizations formed around arts and entertainment.
The field also encompassed organizations involved in hacktivist conflict, associations for firefighters, those related to New
York City prisons and hospitals, among others. As with Level 1, this field includes many organizations local in scope and
excludes matters of national politics. The third level of OWS’s field folds in an additional 59 organizations involved in five
distinct Level 2 fields. These organizations address subjects of education, families and homelessness, local parks and
circuses, the Olympus Scandal, and rock ‘n roll. While the inclusion of these organizations increases the abstraction of
OWS’s field, in many ways they also extend the field’s earlier interests in local institutions, community services, along with
arts and entertainment.
Although these fields may seem quite broad, they exclude a great many organizations. Within TP dataset, the
Louvain method also identified Level 4 fields associated with banking and finance; international affairs; retail, technology,
and telecommunication; basketball; automobiles; and news among others. Within OWS, other Level 3 fields included
organizations in banking and finance, universities and collegiate sports, high art, broadcasting corporations, international
affairs, automobiles, soccer, and, of course, national parties and politics.
5.2 Neighborhood Description
While the Tea Party and Occupy Wall Street have an elevated propensity to appear in articles alongside other organizations
within their same fields, they often coappear in articles with other organizations. Figure 1 illustrates their first order
neighbors in the simplified cofacet network. For visual clarity, we remove TP and OWS as they have direct ties to all
organizations in their respective plots.
These movementcentric plots illustrate pronounced differences between the two movements. First, of the 33
organizations that appear in articles alongside TP, just four organizations have ties only to TP. The majority of TP’s
neighbors coappear in articles with other organizations tied to TP. In contrast, of the 68 organizations mentioned in articles
with Occupy Wall Street, 26 of these organizations never coappear with any of the others,2 suggesting that organizations
within TP’s neighborhood overlap in interests more frequently than those in OWS’s neighborhood.
2
An analysis of these numbers with a chisquared test indicates a significant difference (X2=6.060, p<0.05).
Fig. 1. Movement Neighborhoods
Second, even though the Louvain method parses OWS from national politics, the movement does coappear in
articles with some of these players, however their roles within neighborhood differ greatly from those of TP. Organizations
such as the Democratic Party, the Republican Party, the House of Representatives, and the Senate appear quite central in
TP’s neighborhood. Within OWS’s neighborhood, the role of these state actors is diminished. Further, they form a distinct
cluster, featuring few ties to New Yorkbased organizations, none to New York City arts organizations, yet they do have
relatively more ties to social media outlets.
5.4 Comparing Subgroup Stability
For the following analyses, we test the hypothesis that due to its stateembedded affiliations, TP’s fields and neighborhoods
should demonstrate greater stability than those of OWS. Further, these differences should be most pronounced among the
smaller, more specific fields (Level 1) than the larger, more abstract fields (Level 3). Likewise, lower order neighborhoods
should have greater differences in stability than higher order neighborhoods. As subgroups sizes grow, the stabilizing
influence of stateembedded actors on interorganizational ties should taper as the “rules of the game” become less
understood.
Triadic Clustering. For each subgroup, Table 2 reports the number of closed and open triplets, the global clustering
coefficient, and the effect size expressed as an odds ratio. Supporting our hypotheses, the global clustering coefficient is
significantly greater for TP relative to OWS within the Level 1 fields as well as the first and second order neighborhoods.
The odds of triple closing are approximately 1.5 times more likely for TP than OWS within their Level 1 communities and 3.8
times as likely within their first order neighborhoods. Also supporting our hypotheses, the effect size decreases as the level
of abstraction increases. Exempting TP’s third order neighborhood, its global clustering decreases with community and
neighborhood growth. Curiously, this general drop in TP’s global clustering accompanies an increase for that of OWS such
that its Level 3 field and third order neighborhood are 1.2 and 1.6 times likelier than TP to form a closed triplet.3 Note that
while these two figures are significant, the effect size magnitudes are smaller than those within Level 1 communities and first
order neighborhoods. Given that the effect size diminishes in the proposed direction and that contradictory effects exist only
within the most abstracted subgroups, we maintain that the property of triadic clustering supports both hypotheses.
Table 2. Triadic Closure within Fields and Neighborhoods
Fields Neighborhoods
Level/Order Case Closed Open Cluster Coef Closed Open Cluster Coef
1 TP 702 3005 0.189 594 1005 0.371
OWS 90 580 0.134 360 2332 0.134
2
X 11.232 *** 326.592 ***
OR 1.505 3.827
2 TP 3171 19379 0.141 25572 123625 0.171
OWS 231 1336 0.147 7128 38723 0.155
X2 0.504 63.726 ***
OR 0.946 1.124
3 TP 3180 19605 0.140 77571 358682 0.178
OWS 363 1944 0.157 36954 106434 0.258
2
X 5.317 * 4345.930 ***
OR 0.869 0.623
Coreness. We use two sets tests for coreness stability. The first is a difference of means test on the organizations’ coreness
scores, summarized by Cohen’s d and σ figures to compare effect sizes.4 (See Table 3.) On average, organizations in TP’s
3
We calculate these inversions through exponentiating the absolute value of the log of the reported odds ratios.
4
The degrees of freedom follow Welch’s approximation to avoid equal variance assumptions between cases. The results are
substantively the same using pooled variance estimates.
fields belonged to 3cores while organizations in OWS belonged to 2cores. In terms of their neighborhoods, the
organizations associated with TP typically belonged in cores ranging from four to six and OWS’s neighbors usually
belonged to cores up to three or four k . For all community levels and neighborhoods analyzed, the mean coreness for OWS’s
subnetworks is significantly less than those of TP. Referring to the standardized effects, Cohen’s d and σ, these values
weaken as the fields and neighborhoods grow. These findings strongly support each hypothesis.
Table 3. k Core Means, Medians within Fields (by Level) and Neighborhoods (by Order)
TP OWS Med Diff
Level Mean sd Med Mean sd Med t df d σ 2.5% 97.5% p(TP>OWS)
1 2.804 1.463 3 1.717 1.223 1 5.129 *** 141.142 0.863 0.396 1 2 0.995
2 2.870 1.733 2 1.897 1.032 2 7.909 *** 430.226 0.763 0.356 0 1 0.123
3 2.859 1.726 2 1.966 1.098 2 7.674 *** 570.010 0.643 0.306 0 1 0.082
Order
1 4.324 1.736 5 2.638 1.636 2 4.723 *** 62.397 1.196 0.513 1 4 0.998
2 6.393 3.907 6 3.814 2.232 3 14.562 *** 1056.633 0.896 0.409 2 3 1.000
3 4.858 3.909 4 4.059 3.442 3 7.482 *** 4490.802 0.223 0.111 0 1 0.598
The second set of coreness analyses uses bootstrapped difference of medians tests. As k values are nonnormally
distributed, nonnegative natural numbers with low mean values, nonparametric tests produce more conservative results.
With replacement, we resample the coreness scores for each network and subtract OWS’s median coreness scores from the
median coreness scores produced by TP’s networks. We repeat this procedure 10,000 times to create a distribution of test
estimates. Table 3 reports the observed medians with 95% confidence intervals from these tests along with the proportion of
cases in which the median coreness scores from each of TP’s networks exceed those of OWS.
Within the Level 1 fields the typical coreness of TP’s field affiliates differed by between one and two k values from
OWS’s Level 1 field. This effect did not significantly hold for Levels 2 and 3, though the probability of TP’s median coreness
exceeding OWS’s did decline as the levels became more abstract. Likewise, assessing differences between the
neighborhoods, TP’s first and second order neighbors’ median coreness scores exceeded those of OWS’s by between one
and four k within the first order neighborhoods and between two and three k within the second order neighborhood.
Contrary to the second hypothesis, this effect did not weaken for the second order neighborhood, as the confidence interval
narrows and a slightly greater proportion of its tests (0.2%) match the hypothesized relationship. While the median coreness
score of TP’s third order neighborhood typically exceeds OWS’s (about 60% of the simulations), the proportion of replicated
tests meeting the hypothesis fail to reach the conventional level of 95% statistical certainty.
5.5 Summary of Findings
In general, fields and neighborhoods pertaining to the Tea Party are more stable than those of Occupy Wall Street. The
accuracy of this conclusion largely depends upon the level of abstraction. The most specific communities (Level 1) and the
neighborhoods nearest the analyzed movements (first order) match this empirical statement very closely. The most general
communities (Level 3) and farthest neighborhoods (third order) display either relatively weaker, nonsignificant, or contrary
effects. Less certainty exists regarding the levels in between, as they did not consistently match the hypotheses. The
strongest findingsthat the Tea Party’s closest affiliations are more stable than those of Occupy Wall Street’sholds for
both operationalization of fields as well as pathbased operationalizations of subgroups.
6 Conclusion
This study yields three key implications for movement scholarship. First, the study analyzes two large, interorganizational
networks that include movement and nonmovement organizations alike. From this study, scholars can begin to ascertain
both the diverse array of organizations with whom movements regularly interact, as well as those organizations completely
unaffiliated with and socially distant from movements. Following this contribution, the study furthers implore future research
to examine multiinstitutional politics [i.e., 4]. Further, this work also encourages researchers to assess which fields of
institutional life lack movement organizations and seek structural explanations discerning their absence.
Second, the study approaches the media in a manner slightly different from prior social movements research. Rather
than an entity which covers protests or otherwise brings attention to movement claims, here the media is an outlet that
conveys relationships between organizations and structures fields. Though media coverage entails notorious selectivity
effects [42], typically the coverage accurately portrays the hard facts [17]. Selectivity effects should bear limited
consequences here, as this study analyzes relationship patterns rather than their frequency or representative organizational
coverage. Selectivity effects aside, we have little reason to believe that national media sources systematically mischaracterize
general relationship patterns between two recognized organizations in the same field. That stated, caution should be
exercised in generalizing these findings to universes beyond national news outlets. Afterall, our interaction measurement
only considers whether or not the Times associates two organizations with the same article. While this operationalization
offers an efficient shorthand method to circumscribe fields, more refined interaction codes; such as those indicating allies,
adversaries, or neutral; would provide greater generalizability.
Lastly, this study offers an innovative approach to data collection. As stated earlier, the method offers researchers a
natural update to old techniques of analyzing article synopses. These benefits include offering researchers a customizable
scope, resource efficiency, and accessibility. While this study analyzes only organizational networks during short time
spans, other elements from the data source remain to be explored by movement and organizational scholars. These include
incorporating individuals, locations, and subjects into the analyses. Dynamic organizational network analyses from similar
data sources also prove a subject area ripe for movement research [e.g., 53]. Further, this study presented just one analytic
example which could be conducted from such a data source. While it focused upon two movements, later studies may wish
to examine the fields which circumscribe more movements and nonmovements. Lastly, future research should consider
integrating similar data sources with additional online references.
7 References
1. AlvarezHamelin, José Ignacio, Luca Dall'Asta, Alain Barrat, and Alessandro Vespignani. 2006. “kCore
Decomposition: A Tool for the Visualization of Large Scale Networks.” Advances in Neural Information Processing
Systems 18:41, arXiv:cs/0504107.
2. Amenta, Edwin, Neal Caren, Elizabeth Chiarello, and Yang Su. 2010. “The Political Consequences of Social
Movements.” Annual Review of Sociology 36:287307.
3. Amenta, Edwin, Neal Caren, Sheera Joy Olasky, and James E. Stobaugh. 2009. “All the Movements Fit to Print:
Who, What, When, Where, and Why SMO Families Appeared in the New York Times in the Twentieth Century.”
American Sociological Review 74:63656.
4. Armstrong, Elizabeth A and Mary Bernstein. 2008. “Culture, Power, and Institutions: A MultiInstitutional Politics
Approach to Social Movements.” Sociological Theory 26:1:7499.
5. Bearman, Peter S. and Kevin D. Everett. 1993. “The structure of social protest, 1961–1983.” Social Networks
15:2:171–200.
6. Blondel, Vincent D, JeanLoup Guillaume, Renaud Lambiotte, and Etienne Lefebvre. 2008. “Fast unfolding of
communities in large networks.” Journal of Statistical Mechanics: Theory and Experiment 10:P10008.
7. Bourdeiu, Pierre. 1998. On Television. Translated by Priscilla Parkhurst Ferguson. New York: New Press.
8. CBS New York. 2011. “Groups Plan to ‘Occupy Wall Street,’ But Their Goal Is Not Yet Set.” September 6, 2011.
Retrieved January 19, 2013
(http://newyork.cbslocal.com/2011/09/06/groupsplantooccupywallstreetbuttheirgoalisnotyetset/)
9. CoutureBeil, Alex. 2012. “rjson: JSON for R.” R package version 0.2.9.
10. Csardi, Gabor and Tamas Nepusz. 2006. “The igraph software package for complex network research.” InterJournal,
Complex Systems 1695. http://igraph.sf.net
11. Curtis Jr, Russell L and Louis A Zurcher Jr. 1973. “Stable Resources of Protest Movements: The
MultiOrganizational Field.” Social Forces 52:1:5361.
12. Danon, Leon, Albert DiazGuilera, Jordi Duch, and Alex Arenas. 2005. “Comparing Community Structure
Identification.” Journal of Statistical Mechanics: Theory and Experiment 9:P09008
13. Diani, Mario and Ivano Bison. 2004. “Organizations, Coalitions, and Movements.” Theory and Society 33:281309.
14. Diani, Mario and Katia Pilati. 2011. “Interests, Identities, and Relations: Drawing Boundaries in Civic Organizational
Fields.” Mobilization 16:3:26582.
15. DiMaggio, Paul J. and Walter W. Powell. 1983. “The Iron Cage Revisited: Institutional Isomorphism and Collective
Rationality in Organizational Fields.” American Sociological Review 48:2:147160.
16. Earl, Jennifer and Katrina Kimport. 2008. “The Targets of Online Protest: State and private targets of four online
protest tactics.” Information, Communication and Society 11:4:449–72.
17. Earl, Jennifer, Andrew Martin, John D. McCarthy, and Sarah A. Soule. 2004. “The Use of Newspaper Data in the
Study of Collective Action.” Annual Review of Sociology 30:6580.
18. Fligstein, Neil. 2001. “Social Skill and the Theory of Fields.” Sociological Theory 19:2:105–25.
19. Fligstein, Neil and Doug McAdam. 2011. “Toward a General Theory of Strategic Action Fields.” Sociological
Theory 29:1:126.
20. Fligstein, Neil and Doug McAdam. 2012. A Theory of Fields: Toward a Theory of Strategic Action Fields. New
York: Oxford University Press.
21. Fogarty, Jim. “Occupy Wall Street Protests Reject US Institutions” Epoch Times, October 11, 2011. Retrieved
January 19, 2013
http://www.theepochtimes.com/n2/opinion/occupywallstreetprotestsrejectusinstitutions62696.html
22. Friedkin, Noah E. 1984. “Structural Cohesion and Equivalence Explanations of Social Homogeneity.” Sociological
Methods and Research 12:3:23561.
23. Friedkin, Noah E. 2004. “Social Cohesion.” Annual Review of Sociology 30:40925.
24. Granovetter, Mark S. 1973. "The strength of weak ties." American Journal of Sociology 78:6:13601380.
25. Gross, Neal and William E. Martin. 1952. “On group cohesiveness.” American Journal of Sociology 57:546–54.
26. Hardt, Michael and Antonio Negri. 2011. “The Fight for ‘Real Democracy’ at the Heart of Occupy Wall Street.”
Foreign Affairs October 11, 2011.
27. Holland, Paul W and Samuel Leinhardt. 1972. “Some Evidence on the Transitivity of Positive Interpersonal
Sentiment.” American Journal of Sociology 77:6:12051209.
28. Johnson, Erik W., Jon Agnone, and John D. McCarthy. 2010. “Movement Organizations, Synergistic Tactics and
Environmental Public Policy.” Social Forces 88:5:226792.
29. King, Brayden G and Nicholas A. Pearce. 2010. “The Contentiousness of Markets: Politics, Social Movements, and
Institutional Change in Markets.” Annual Review of Sociology 36:24967.
30. Krackhardt, David. 1999. “The Ties that Torture: Simmelian Tie Analysis in Organizations.” Research in the
Sociology of Organizations 16:183210.
31. Lang, Duncan Temple. 2012. “RCurl: General network (HTTP/FTP/...) client interface for R.” R package version
1.911.
32. Lind, Benjamin and Judith StepanNorris. 2011. “The Relationality of Movements: Movement and
Countermovement Resources, Infrastructure, and Leadership in the Los Angeles Tenants’ Rights Mobilization,
1976–1979.” American Journal of Sociology 116:5:1564609.
33. McAdam, Doug. 1982. Political Process and the Development of Black Insurgency 19301970. Chicago:
University of Chicago Press.
34. Minkoff, Debra and John D McCarthy. 2005. “Reinvigorating the Study of Organizational Processes in Social
Movements.” Mobilization 10:2:289308.
35. Newman, Mark EJ. 2006. “Modularity and community structure in networks.” Proceedings of the National Academy
of Sciences, USA 103:23:85778696.
36. Newman, Mark EJ and Michelle Girvan. 2004. “Finding and Evaluating Community Structure in Networks.” Physical
Review E 69:026113.
37. New York Times. 19481970. The New York Times Index.
38. New York Times. 2012a. “Article Search API.” http://developer.nytimes.com/docs/read/article_search_api . Last
accessed December 7, 2012.
39. New York Times. 2012b. “The TimesTags API.” http://developer.nytimes.com/docs/read/timestags_api . Last
accessed December 11, 2012.
40. O’Hara, John M. 2010. A New American Tea Party. Hoboken, NJ: Wiley and Sons.
41. Osa, Maryjane. 2003. Solidarity and Contention: Networks of Polish Opposition. University of Minnesota Press.
42. Oliver, Pamela E. and Daniel J. Myers. 1999. “How Events Enter the Public Sphere: Conflict, Location, and
Sponsorship in Local Newspaper Coverage of Public Events.” American Journal of Sociology 105:3887.
43. Opsahl, Tore. 2013. “Triadic closure in twomode networks: Redefining the global and local clustering coefficients.”
Social Networks 35, doi: 10.1016/j.socnet.2011.07.001
44. Powell, Walter W, Douglas R White, Kenneth W. Koput, and Jason OwenSmith. 2005. “Network Dynamics and
Field Evolution: The Growth of Interorganizational Collaboration in the Life Sciences.” American Journal of
Sociology 110:4:1132205.
45. R Core Team. 2012. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical
Computing.
46. Redmond, Tim. “The Goals of Occupy Wall Street” San Francisco Bay Guardian, October 5, 2011. Retrieved Jan 19,
2013 http://www.sfbg.com/politics/2011/10/05/goalsoccupywallstreet
47. Rohlinger, Deana A. 2007. “American Media and Deliberative Democratic Processes.” Sociological Theory
25:2:122148.
48. Rucht, Dieter. 2004. “Movement Allies, Adversaries, and Third Parties.” In David A. Snow, Sarah A. Soule, and
Hanspeter Kriesi (eds.), The Blackwell Companion to Social Movements. Malden, MA: Blackwell Publishing, p.
197216.
49. Seidman, Stephen B. 1983. "Network structure and minimum degree." Social Networks 5:3:269–87.
50. Tea Party Patriots. “About.” Retrieved: January 18, 2013 http://www.teapartypatriots.org/about/
51. Wasserman, Stanley and Katherine Faust. 1994. Social Network Analysis: Methods and Applications. New York:
Cambridge University Press.
52. Watts, Duncan J and Steven Strogatz. 1998. "Collective Dynamics of 'SmallWorld' Networks." Nature
393:6684:440–442.
53. Wang, Dan J and Sarah A. Soule. 2012. "Social Movement Organizational Collaboration: Networks of Learning and
the Diffusion of Protest Tactics, 19601995." American Journal of Sociology 117:6:16741722.
54. Stanley, Stephanie. 2012. The Tea Party and Occupy Wall Street Movements: Populism and Protest. Portland State
University: University Honors Thesis.
55. Zernike, Kate. 2010. Boiling Mad. New York, NY: Times Books.