The Dynamics of Collective Violence:

Dissecting Food Riots in Contemporary Argentina

Javier Auyero, State University of New York-Stony Brook
Timothy Patrick Moran, State University of New York-Stony Brook

This article combines a statistical analysis with qualitative research to

investigate the dynamics of collective violence in one of its most recurrent
forms - the food riot. Using an original dataset collected by the authors
on 289 food riot episodes occurring in Argentina in December 2001,
the article argues for the need to dissect the local, contextualized inner-
dynamics of the episodes. We find significant interrelationships between
three important factors: the presence or absence of police, the presence
or absence ofpolitical party brokers, and the type of market looted (big/
chain or small/local). We then conduct a qualitative and ethnographic
analysis to illustrate how these interactions might play out in two ideal
type looting scenes -one illustrating the role of public authorities at big,
chain supermarkets, the other showing the importance of party brokers
at small, local food markets. We conclude by calling for more such
research to better understand the mechanisms and processes, especially
the relationship between state power and party politics, involved with
all forms of collective violence.

Introduction
On the morning of Dec. 14, 2001, in Rosario, Argentina, groups of poor people
gathered in front of supermarkets demanding food. When denied, they began
breaking into the premises and taking away merchandise. Store owners,
managers, and employees, looked on in bewildered disbelief, and policemen,
if reaching the stores in time, attempted to prevent further looting either by
negotiation or force. During the next few days, the sacking of grocery stores
and supermarkets extended unevenly throughout the country, and by the end of
the week of December 14-21, episodes of food lootings occurred in 11 of the 24
Argentine states. Close to 300 stores were looted. Eighteen people (all of them

We wish to thank Mia Bloom, Daniel Pridman, Leslie Gates, James Jasper, Jackie Klopp,
John Krinsky, Roy Licklider, Francesca Polletta, Sherrill Stroschein, Sidney Tarrow and
Charles Tilly for their helpful comments to a previous draft of this paper during a lively
session at the Columbia Contentious Politics Seminar. We also presented a draft of this
paper at the Economic Sociology Workshop at Princeton University; many thanks to
Patricia Fernandez-Kelly and Viviana Zeiizer for a constructive session. We also thank
for criticisms and suggestions Gaston Beltrdn and John Markoff; our colleagues at Stony
Brook, Michael Schwartz, Naomi Rosenthal, Andrea Tyree, Ian Roxborough; and three
anonymous reviewers. Funding for this research was provided by the Harry Frank
Guggenheim Foundation (PI: Javier Auyero). Direct correspondence to either author at
Department of Sociology, SUNY-Stony Brook, Stony Brook, NY 11794-4356. E-mail:
[email protected] or [email protected].
® The University of North Caroiina Press Social Forces. Voiume 85, Number 3, March 2007
1342 • SocialForces Volume85,Number3 • March2007

under 35 years old) had been killed, either by the police or by store owners.
Hundreds more were seriously injured, and thousands were arrested.
Employing a research design that combines a statistical analysis with qualitative
and ethnographic research, this research provides an empirical examination of
the wave of food riots. Using an original dataset on 289 looting episodes, we
describe and explain the diversity and dynamics of collective violence by paying
empirical attention to two key interacting dimensions: 1.) the role of political
entrepreneurs in the promotion, inhibition and/or channeling of physical damage
to objects and persons; and 2.) the impact that policing has on the development
and form of transgressive contention.
Our data collection was guided by existing research and theoretical perspectives
on these two dimensions of collective violence. First, scholarship in the field is
increasingly recognizing and analyzing the role of "political entrepreneurs" (and
the networks they mobilize) in the promotion of damage-making joint actions.
Historians, political scientists, anthropologists and sociologists (Wilkinson 2004;
Roldan 2003; Tilly 2003; Blok 2001; Kakar 1996; Das 1990; Shaheed 1990)
acknowledge the role of what a recent report on the social causes of violence
published by the American Sociological Association calls "third parties" (Levine
and Rosich n/d:70). According to the report, these third parties are "often involved
or present during violent encounters; yet, our knowledge of their role is very
limited." (For an exception, see the Marx 1974 article on the role of the "agent
provocateur" and the "informant" in the origins and course of collective action.)
Our research on the Argentine 2001 lootings seeks to expand our still partial
knowledge of the role of "third parties" (in our case, political party brokers) in the
makings of collective violence.
The second set of theoretical ideas and scholarly debates that guided the
dataset revolve around policing. There is a substantial body of work that examines
the ways in which state repression affects collective action (Davenport 2000;
Fantasia 1988; Gould 1995; Hirsch 1990; Piven and Cloward 1977; Zhao 2001;
more recently, Davenport, Johnston and Mueller 2005). Repression is a key
explanatory variable in many studies of movement mobilization and tactic adoption
(della Porta 1995, 1996; Ellison and Martin 2000; McAdam 1982, 1988).' Most of
the literature, however, concentrates on the consequences of protest policing on
a particular type of collective action-i.e., social movements (for an exception, see
Salert and Sprague 1980). The effects of policing during food riots have thus far
not been examined. By looking at the difference repression made in the selection
of the crowd's targets during an intense wave of looting, we seek to refine current
understandings of the influence of police tactics on transgressive contention.
Riots - both race and ethnic ones - have been widely researched in U.S.
scholarship, mainly focusing on the 1960s' wave and on the 1992 episodes
in Los Angeles (see Stark et al. 1974; Baldasssare 1994; for comprehensive
reviews see McPhail and Wohlstein 1983 and Useem 1998). There are plenty
of now-classic studies on the individual attributes of participants in riots
(Caplan 1970; Caplan and Paige 1968; Moinat et al. 1972), as well as classic
and contemporary studies on the demographic, economic, ethnic and racial
composition of rioting communities (Spillerman 1970, Lieberson and Silverman
 Dynamics of Collective Violence '1343

1965; Wohlenberg 1982; Bergensen and Herman 1998). One of the strengths
of this U.S.-based scholarship is the emphasis on the complexity, diversity,
interactive and dynamic character of lootings.
We seek to apply these insights to apply these insights to a phenomenon
heretofore unexamined. In particular, we draw upon two key insights of this body
of literature: 1.) the relational underpinnings of lootings: Riots are carried out in
small groups of people who are connected in some ways (through friendship,
family, and/or community ties) and assemble, remain and disperse together
(Dynes and Quarantelli 1968; Aveni 1973; McPhail and Wohlstein 1983); and 2.)
the selectivity of looters' actions: Farfrom being random collective actions, looters
selectively target particular kinds of stores (based on the ethnicity of store owner,
on the type of store and/or other variables) (Rosenfeld 1997; Greenberg 1992).
However, with the notable exceptions of the seminal studies by Walton and
Seddon (1994) and Walton and Shefner (1994), food riots in Latin America have not
been examined with the same degree of theoretical sophistication and empirical
rigor- despite being still widespread in the region.^ Walton and Seddon (1994)
analyze the wave of popular protest that followed the implementation of structural
adjustment policies and government austerity measures in the developing world
from the 1970s to the 1990s (for a more general statement see Walton and Ragin
1990). These authors contend that the specific origin of "austerity protests" (i.e.,
"large-scale collective actions including political demonstrations, general strikes,
and riots" [1994;39]) lies in the period of global adjustment that ensued the
international debt crisis. As they put it:

"The broad implementation of austerity measures as a

condition of structural adjustment and debt restructuring
represented an attack on the very means that made urban
life sustainable. Austerity led to popular protest in the times
and places that combined economic hardship, external
adjustment demands, hyperurbanization, and local traditions
of political mobilization." (1994, p. 99, our emphasis)

Our paper seeks to complement previous research by 1.) examining the internal
dynamics of riot episodes rather than their causes (a task actually hinted at when
these authors gather case based evidence [see specifically, Walton 1989]); and
2.) scrutinizing the political dimensions of rioting (an important complement to
the more structural analysis).
Europeanist social historians who study collective responses to 18th century
subsistence crises share an interest in the political aspects of rioting. In his
study of riots in England and Wales at the turn of the 18th century, for example,
Bohstedt (1983) asserts that we should look at the strength and stability of both
horizontal networks (relations between members of a community, based on
kinship, market, neighborhood) and vertical networks (relationship between that
community and the elites and authorities) in order to understand and explain the
episodes. Riots, in Bohstedt's view, were expressions of community politics.
Noting the ambiguous role played by local authorities and notables, Cynthia
1344 • Socifl/Forces Volume 85, Number 3 « March 2007

Bouton's (1993) examination of the French Flour War also highlights the political
side of food riots.^ Bouton (1993), Thompson (1994) and Markoff (1996) also bring
to light an aspect of the 18th century European subsistence riots that will prove
key to understand and explain the dynamics of the 2001 lootings in Argentina:
the close and crucial relationship between the workings of patronage networks
and the development and outcome of riots, between established ways of doing
politics and extra-ordinary ways of expressing collective suffering.
The first section of the article describes the episodes and positions them
in a broader, 20-year wave of protest. Then, after a description of the archival
and ethnographic fieldwork and the contents of the dataset, we examine
two distinct patterns of observed looting - looting occurring in large, chain
supermarkets vs. looting taking place in smaller, locally owned stores. We
offer explanations as to the inner dynamics of these two patterns by focusing
on the relationships between the type of market looted, police presence
and the presence of political party brokers. Drawing on qualitative analyses
of newspaper and investigative journalists' reports as well as ethnographic
fieldwork on Argentine political networks, we then create likely narratives to
illustrate how the statistical relationships revealed in the dataset might actually
play out in different riot settings. The article concludes with speculation about
the mechanisms in play during these lootings.
Our findings indicate distinct types of "mini-lootings" discernible within what
has been previously presented as a single, large-scale event. Far from being
monolithic phenomena, we argue, food riots are relational processes in which
local context matters, particularly the actions of political brokers and police
officers. Attention to these two key actors shows that far from being a break
with regular politics, collective violence should be seen more as a continuum
with everyday forms of political activity.

Contentious Argentina
Although spectacular in their intensity and extent, the lootings of December 2001
were hardly isolated moments of collective violence in Argentina. They were a
peak within a "cycle of collective action." (Tarrow 1998) During the previous two
decades, new and unconventional forms of popular contention transformed
Argentina into a veritable landscape of violent protest. Sieges of (and attacks on)
public buildings (government houses, legislatures, courthouses), barricades on
national and provincial roads, and sit-ins in central plazas became widespread
throughout the country. Though fewer in number of episodes, participants and
amount of material and human damage than those of December 2001, lootings
of markets and food stores also took place in May through July of 1989 and
on March and June 1990 throughout the country. Most scholars (Svampa and
Pereyra 2003; Villalon 2002; Giarraca 2001) point to deproletarianization, state-
retrenchment and decentralization of state services as the processes that lie at
the root of the upsurge of contention.
The burst of food riots in December 2001 brought out thousands of people
who blockaded roads and bridges throughout the country, banged pots and
pans in the main plaza of Buenos Aires (the episodes known as cacerolazos]
 Dynamics of Collective Violence '1345

and ousted two presidents in less than a month - indicating that a historical shift
in claims-making was still under way. The peak of collective action during 2001
has been the subject of many scholarly (Cafassi 2002; Fradkin 2002; Lewkowicz
2002), journalistic (Bonasso 2002; Martinez 2002; Camarasa 2002), and insiders'
accounts (Kohan 2002; Colectivo Situaciones 2002). Central as they were in
provoking a terminal political crisis, the Argentine lootings, however, remain an
uncharted terrain for social scientists.'' The few existing reports on the 2001 food
riots are single-actor accounts dominated by what sociologist Charles Tilly (2003)
calls the "steam boiler" analogy or what historian E.P Thompson (1993) labels a
"spasmodic view" of popular revolt. In other words, the main actors in the lootings
are said to be the poor and unemployed who, responding to a rapid reduction
in the standard of living by visible government (in)action (e.g., the suspension of
many food distribution programs) and the high level of joblessness (in December
unemployment rates were 21 percent of the economically active population),
suddenly exploded in anger and plundered stores and supermarkets (Fradkin
2002). As analysts in Argentina portray the events, poverty and unemployment,
together with state inaction, created an insurmountable pressure that built up
during 2001 until everything exploded. Mosttellingly, the titleof a widely publicized
book on the 2001 events is Olla a Presion or "pressure cooker." (Cafassi 2002)
Yet these prevailing analyses do not account for what even a superficial
observation of the recent lootings would find: as the literature on food riots
in other parts of the world clearly shows (Walton 1989; Walton and Seddon
1994), the looting episodes had importantly different inner dynamics. Attacks
on supermarkets and stores had different degrees of mass participation and
police action, and the targets of the looting crowds were also diverse. After
describing our methodology in detail, we delve into these contextual factors
more closely in section three.

Data and Methodology

We constructed a database of these riots based on an exhaustive analysis
of journalistic accounts of the week's events. There is a long tradition in the
collective action literature demonstrating the usefulness of newspaper archives
for the collection of event data (Earl et al. 2003; Kriesi et al 1995; McAdam
1982; Olzak 1989, 1992; Tarrow 1989; Tilly 1995) and we are hardly the first
to use newspapers data to study rioting (Danzger 1975; Walton and Ragin
1990; Walton and Shefner 1994; Walton and Seddon 1994). We are, however,
aware of the problems and controversies surrounding the use of newspapers
as a source of data on contentious events (Franzosi 1987; Koopmans and Rucht
1999, 2002; Myers and Caniglia 2004).^ We relied on many sources of data (e.g.,
Bessinger 1998; Myers 1997). Specifically the data are drawn from three national
newspapers with different political leanings - from left to right Paginal2, La
Nacion, and Clarin - and the main local newspapers of the ten provinces where
lootings occurred - El Ciudadano, La Voz del Interior, La Mariana del Sur, Rio
Negro, Cronica-Chubut, La Gaceta, El Litoral, El Liberal, Los Andes, and El Sol.
Because all the lootings took place during one week, we were able to read every
print and on-line edition of the 13 newspapers for the week, thus providing an
 1346 • SocialForces Volume85,Number3 • March2007

exhaustive dataset that avoids the problems related to sampling techniques (Earl
et al. 2004:68). This study relies mainly on "hard news items" (i.e., the who, what,
when and where of the episodes) which, as other researchers have pointed out,
are generally more precise than "soft news." (i.e., journalists' impressions and
inferences) (Earl et al. 2004:72) We are thus confident about the accuracy and
reliability of the journalistic accounts in this context.^
For the purposes of data collection, a looting episode is defined as the activity
of two or more persons who 1.) forcibly seize objects over restraint or resistance,
or 2.) attempt to seize objects but are effectively met with restraint or resistance.
Some groups at the markets demanded food without attempting to force
entrance. These episodes are labeled "claims for food." Thus, we coded all 289
riot episodes, representing every looting, attempted looting, or claim for food
reported by the Argentine press to have happened the week of December 14-21.
Of these, 261 are successful lootings or attempted lootings rebuffed by force
(hereafter simply lootings), and 28 are collective claims for food. Each episode
is a discrete event though they occurred in one continuous riot week. For each
episode we recorded the following information when it was available:

1.) Location of Episode: Coded at four levels of analysis (province, statistical

metropolitan area (SMA, caWed aglomerados in Argentina), city, neighborhood).

2.) Number of Participants (estimated): Coded as a four-category ordinal

variable (less than 100, between 100 and 400, between 400 and 600, more than
600) based on either the numerical estimates or the descriptive language used
by the reporter.

3.) Type of Market Looted: Partitioned into three categories: a) foreign-

owned chain superstores (known as hipermercados), or other nationally- or
provincially-owned chain supermarkets, b) small, locally-owned supermarkets or
neighborhood grocery stores, c) non-food sites such as pharmacies, shoe stores,
bicycle shops or public buildings.

4.) Presence/Absence of Political Party Brokers: The Partido Justicialista (or

Peronist Party) remains deeply embedded in Argentine society, particularly in
working-and lower-class neighborhoods where local Peronistorganizations(called
"base units") rely on extensive informal and clientelistic networks to distribute
resources and recruit support (see Auyero 2001; Levitsky 2003). A key actor in
these social networks is the local puntero (translated as "neighborhood broker").
Punteros mobilize the neighborhood for elections and other political activities,
and are in charge of the distribution of state welfare benefits (such as food,
medicine, and unemployment subsidies). Evidence of the presence of brokers is
based on the descriptive language of reporters. Specifically, this variable is coded
"yes" when reporters speak of "punteros" or "dirigentes barriales Peronistas"
(another term for Peronist brokers). When these terms were unspecified or when
language was inconclusive, the variable is coded "no."'
 Dynamics of Collective Violence '1347

5.) Presence/Absence of Police: In Argentina, police forces are either funded

by the federal government (Policia Pederal) or by the provinces (Policia Provincial).
The Federal Police operates within the territory of the Federal Capital and outside
of it when crimes are considered under federal jurisdiction. Provincial-police
operate within the boundaries of each of the 23 states. During the lootings, both
federal and state police were active at the looting scenes. If either were reported
to be present, or if arrested were reported (marking their presence indirectly),
this variable is coded "yes." As with brokers, when the presence of police was
unspecified or when language was inconclusive, the variable is coded "no."

Q.) Additional Variables: we also recorded information when provided on the

number of people arrested (17 episodes), injured (22 episodes) or killed (18
episodes), and whether or not any distinguishing characteristics of the rioting
group were reported. These events were either too infrequent, or the data proved
much less available, to be included in the final analysis. The day and time of the
episodes was also noted. No significant relationship between these two variables
and the riot's size and/or location was found.

The qualitative and ethnographic portion of this analysis is drawn primarily

from three sources: 1.) The 128 episodes for which detailed, investigative
reporting exists, allowing a more in-depth analysis of patterns of looting activity,
especially the seemingly minor incidents that triggered violence and the presence
of organizers; 2.) Ethnographic fieldwork on political networks carried out in
Buenos Aires, Neuquen, and Santiago del Estero during 1995-1997 and 2000-
2001 ;^ and 3.) Secondary literature on grassroots political networks in Argentina
(Torres 2002; Levitsky 2003; Goldberg 2003).

Dissecting Food Riots

The descriptive analysis of the looting episodes showed much diversity in
terms of their location, number of participants, type of store attacked, and
the presence of police and party brokers among the crowds. The food riots
were unevenly distributed in broad geographic terms. A third of the recorded
episodes occurred in Buenos Aires (96), the most populated province. Another
20 percent occurred in Santa Fe (61), the third most populated province. About
10 percent occurred in each of the two southern states, Neuquen (29) and
Rio Negro (27), and the northern state of Tucuman (27), and the remaining 49
episodes were scattered over seven other provinces. Eleven provinces had no
reported episodes during the week. An analysis of the provincial spread of
the riots through time shows a random diffusion of looting activity, leading
us to conclude that they are not manifestations of either a "diffusion-of-an-
innovation" or of "collective contagion" of the kind detected in the urban riots
that erupted in many American cities during the 1960s.
Locally, the crowds attacked different types of targets as well. Nearly 60 percent
of the episodes took place in small, local markets and grocery stores (Category
2), close to a third occurred in big, chain-owned supermarkets (Category 1),
and 8 percent occurred at non-food sites (Category 3). Successful lootings and
 1348 • SocialForces Volume85,Number3 • March2007

those effectively stopped by restraint were evenly distributed when the targets
where Category 1 supermarkets, yet in the case of episodes at Categon/ 2 sites,
successful lootings outnumbered rebuffed attempts by nearly 3 to 1.
There was also variation in the number and type of actors involved in the
riot episodes. The number of participants ranged from the thousands in
episodes occurring in Concordia (Entre Rios), Banda del Rio Salf (Tucuman),
and Centenario (Neuquen), to the dozens in many smaller episodes in Rosario
(Santa Fe), Guaymallen (Mendoza), and Parana (Entre Rios). Of the episodes with
count information, the modal category of estimated participants was between
100 and 400 (close to 70 percent). Police presence was reported in 106 of the
289 episodes (37 percent), sometimes outnumbered by the rioters, other times
not, sometimes deterring the crowds with rubber bullets and tear gas (and in a
few reported cases real bullets), other times dissuading potential looters simply
by their presence. The number of arrests also varied widely from province to
province, from dozens in Entre Rfos, to approximately 200 in Rosario, to close to
600 in Tucuman.^ The presence of political party brokers was reported in 67 of
the episodes (23 percent).

The Inner-Dynamics of Food Riots

One of the goals of this research is to examine the riots not as single, monolithic
phenomena, but rather as socio-political events with recognizable internal
differences (see Walton 1989; McPhail and Wohlstein 1983).^° More specifically,
we examine episodic data to investigate a set of relationships that no study of
lootings and/or food riots have thus far examined, i.e. the links between the
looting site itself, and the presence of police and/or political brokers at the site.
Table 1 presents the frequency distribution of the reported presence of brokers
and police at the looting site by type of market looted. Of the 289 total episodes
in the dataset, the type of market looted was not determinable in five of the
episodes. Another 23 occurred in non-food markets such as public buildings that
stored food products for state-funded programs, or pharmacies, shoe stores, etc.
To concentrate on a more homogeneous target site, these episodes are excluded,
leaving a subsample of 261. Political brokers were reported at 45 (25 percent) of
the looting episodes, and police at 94 (36 percent). Interestingly, however, their

Tahle 1: Frequency Distribution of 261 Riot Episodes hy Market Type

Presence of Brokers Presence of Police

Market Type Yes No Yes No
Large, Chain Supermarket 13 79 53 39
Column percent (19.7) (40.5) (56.4) (23.4)
Row percent (14.1) (85.9) (57.6) (42.4)
Small, Local store 53 116 41 128
Column percent (80.3) (59.5) (43.6) (76.6)
Row percent (31.4) (68.6) (24.3) (75.7)
 Dynamics of Collective Violence '1349

presence at the two types of markets differed. For brokers, the column percents
are most telling. When brokers are present at the looting, the site is usually a
small, local store (Category 2) - 80 percent of the time brokers were reported
at such sites, compared to 20 percent at large, chain supermarkets (Category
1). In the absence of brokers, however, the two market types were looted at
closer rates. The opposite is true for police, where the row percents are more
interesting. In terms of police presence at looting sites, type of market does not
seem to make much of a difference, but police non-response was overwhelmingly
at small, locally owned stores.
To more formally assess these differences. Table 2 presents the results of a
logistic regression analysis predicting the type of market looted. Market type is
a dummy variable where 1 = Category 1 looting site, and 0 = Category 2 looting
site (see methods section for descriptions), thus the signs of the coefficients
should be interpreted with respect to large, chain supermarkets. As anticipated
by the frequency distribution. Model 1 shows the strong, statistically significant
effect of broker presence in small, local market looting. Subtracting one from
the odds ratio we see that the odds of broker presence at a looting site are 64
percent lower when the site is a Category 1 chain supermarket.
But the relationship between brokers and looting site is more complex than
brokers simply preferring to direct their followers to small, locally owned markets.
The presence of police is added in Model 2, bringing two interesting effects. First,
the overall fit of the model is much better. Police presence is an even stronger
predictor of market type looted - the odds of police presence at a looting site
are 268 percent higher when the site is a category one, large chain supermarket.
Second, the effect of broker presence is substantially reduced to the point of
losing statistical significance. The odds of broker presence at a looting site are
still lower when the site is a Category 1 chain supermarket, but this effect is no
longer statistically significant when controlling for the presence of police. Thus
while police response is not affected by the presence or absence of brokers, the
reverse is not the case. While the frequency distribution and the results of Model
1 show that brokers prefer to direct their followers to small, locally owned stores
as opposed to large, chain supermarkets, they are doing so largely because the
police are not there. Once police presence is held constant across the looting
episodes, brokers have no reason to prefer one type of market over another. In
short, brokers do not prefer smaller markets they prefer safer ones.
To extend the analysis further. Model 3 addresses the relationship between the
size of the crowds and the presence of brokers and police. Intuitively one might
expect the number of people involved in the lootings to also be a an important
factor in determining the effects of broker and police presence - brokers should
be able to gather more people through organization, and hundreds of people
looting a store might warrant more police action than tens of people. Yet in Model
3, this variable has altogether no effect on either 1.) the overall fit of the model,
or 2.) the size of the predictive effects of market type and broker presence. Police
are less likely to respond to lootings at small, locally owned stores regardless of
whether brokers are present or the size of the rioting crowd.
1350 • SocialForces Volume85,Number 3 • March2007

The fact that police responded less frequently to lootings at small markets
concurs with reports that the state and federal police took special care when
it came to protecting stores like the French-owned Carrefour or the American-
owned Norte, while at the same time creating what grassroots activists called a
"liberated zone" around small and medium-size stores, allowing political brokers
and crowds to move freely from one target to the next. "The police acted very
well," said an owner of one of the largest supermarket chains in Rosario, "Police
action was impeccable." Juan Milito, the head of Rosario's Shopkeepers Union
(Union de Almaceneros or "Small Stores") said, "As always, the small stores were
the most damaged... the big chains were protected." (El Ciudadano 12/24/01)
A highly perceptive observer of police behavior (and former undersecretary of

Ulit CO CO CN
0) CO
CD in
CO
O c to •<-^
lei

o
Coeff.
Log.

OO
0)
CO
CN
O o CO
"55
oeff.

03 CO CU (32 T- CO CO
o
o
lO CO CO CSI
s CN in •»—
r
CD CO CO
CO CN
eff.
Ids

CO

o oo
CO

"55
•a
o
. it CN CO lO CO
a> C3 CO 1— r-.-
o o6 d T- ^ ES
O
CN T— CO
CO CN

>- o == Si « S
II II

n CD
Q.
re CD 'o
r~ O
8
Pre sei
ent

a> (2
olic Pres

stin ted

•o
c O V .S
S. k_ CD
is
ons
apui

o
oj p ^ C
n LU
o
Z; Z m pq
 Dynamics of Collective Violence • 1351

security in the state of Buenos Aires), speaks of "evident police passivity" during
the lootings, and adds that it is quite common for the state police to "liberate"
or free zones so that criminal activity can proceed without police presence (Sain
2004). As the lootings were taking place, journalist reports from several media
noted that the Buenos Aires police was "not very disposed to intervene" (Kollman
12/20/01); the National Guard (Gendarmeria), the Federal Police, and the Naval
Police (Prefectura) had a similar diagnosis.
The importance of political party brokers also coincides with two investigative
journalist reports describing their organizing and coordinating activities before
and during the food riots. Flyers circulated throughout poor neighborhoods in
Greater Buenos Aires, for example, inviting residents to join the looting crowds:
"You are invited to destroy the Kin supermarket this coming Wednesday at
11:30 a.m., the Valencia supermarket at 1:30 p.m., and the Chivo supermarket
at 5 p.m." read one. These flyers were distributed by members of the then-
oppositional Peronist Party, some of them local officials, others well-known
grassroots leaders." In the dataset presence of brokers was reported in only 25
percent of the episodes, yet in the 128 episodes in which there exists detailed
reporting, 52 percent reported broker presence, suggesting that their presence
may be an even greater factor than uncovered here.
Evaluating Model 2 for each of the four combinations hints at a more specific
understanding of broker and police presence. Based on the model, the expected
probability of small, local markets being the looted site is lowest when police are
present and brokers are not (0.42) and highest when brokers are present and police
are not (0.83) - a difference of 0.41. The difference in expected probabilities when
brokers and police are both present (0.57) and are both absent (0.73) is only 0.16.
These findings are confirmed by a more disaggregated look at the frequency
distribution from Table 1 (Table 3). First, the relationship between police
presence and broker presence over all 261 episodes leads one to conclude that
this relationship is statistically significant (chi-square is 24.75), and that while
the absence of brokers leads to a roughly equal chance of police presence,
the presence of brokers almost always is associated with no police (almost 90
percent of the time). But looking at the next two cross-tabs, we see how the
broker and police presence connection operates through the type of market
looted, and more specifically whether or not the looting site is a small, local
market. When the site is a big, chain market, there is higher likelihood of police
presence than not, but there is little broker involvement in these episodes, and
the chi-square value (.08) suggests that the episodes are more or less randomly
distributed across the four cells. But when the site is a small, local market, we
see much more broker activity and a much lower likelihood of police presence.
In every episode occurring in these markets, there was never a riot in which
brokers and police were both present - all 53 episodes in which brokers were
involved, police were not.
Together then, these findings point to two important contextual dimensions of
food rioting, connections that analysts of the recent wave of violent contention
in Argentina have consistently overlooked. First, in terms of brokers, there is
significance in the often obscure (and obscured) links that looters maintain with
1352 • SocialForces Volume85,Number 3 • March 2007

Table 3: Frequency Distributions of Police and broker Presence by Market Type

All Incidents (n = 261)
Presence of Brokers
Presence of Police Yes No
Yes 7 87
(10.6) (44.6)
No 59 108
(89.4) (55.4)
Chi-Square = 24.75 ***

Large, Chain Supermarkets (n=92)

Presence of Brokers
Presence of Poiice Yes No
Yes 7 46
(58.9) (58.2)
No 6 33
(46.2) (41.8)
Chi-Square = .08

Smaii, Local iVIarkets (n = 169)

Presence of Brokers
Presence of Poiice Yes No
Yes 0 41
(0.0) (35.3)
No 53 75
(100.0) (64.7)
Notes : ***p < .001
Numbers in parentheses are column percents

political entrepreneurs and, through them, with established power-holders,

as upheld by previous research in other parts of the world. Research on the
origins and forms of communal violence in Southeast Asia (Wilkinson 2004;
Kakar 1996; Shaheed 1990), for example, show that violent entrepreneurs enjoy
important links with established political parties - or as Veena Das (1990:12)
puts it, with the "underlife of political parties." In Latin America, a recent study
of "la violencia" in Antioqufa, Colombia (Roldan 2002) also highlights the role
played by partisan politics and state agents in the promotion of collective
violence (for Colombia, see also Braun 1980 and Sanchez and Meertens 2001;
for ethnographic descriptions of the links between party politics and violence
in Brazil see. Arias 2004 and Goldstein 2003).
 Dynamics of Collective Violence • 1353

The second contextual dimension regards the crucial disparities in the

distribution of state repressive actions during the lootings (itself contingent upon
the differential distribution of economic and political power among actors, in
this case store owners). Findings point to the decisive impact that policing has
on the development and form of collective violence. Although the phenomenon
under investigation here is episodic contention and not social movement activity,
this second set of findings dovetails with current research on the relationship
between state repression and popular contention (Davenport, Johnston and
Mueller 2005). More specifically, this analysis supports the claim that the level of
threat posed by a looting event is intimately associated with police intervention
(Tilly 1978; McAdam 1982; Davenport 2000), While the findings support the
theoretical importance of threat in these processes, it is important to note that we
conceptualize this term somewhat differently than in the current literature (see
Earl et al, 2003), For us, "threat" is not defined in terms of number of participants
in a looting (found not to be related to police presence), the use of confrontational
tactics (constant during the week under investigation), or the adoption of radical
goals (absent among the looters), but in terms of the type of market looted and
its association with powerful economic and political interests.
On the other hand, our findings contradict another strand of the repression
literature that suggests that the "weakness" of movements (i,e,, those composed
of subordinated groups) is strongly correlated with police repression (Gamson
[1975] 1990; Piven and Cloward 1977), In this research, police presence is not
linked so much with the composition of the crowd (which always entailed the
poor and the unemployed), but with the type of target selected for looting and
with the presence or absence of "third parties" among the looters.

Two Looting Scenes

In order to begin an examination of the mechanisms whereby the relationships
found in Section 3 play out before and during the riot, we construct two ideal-
type "looting scenes" (one taking place in big supermarkets and the other
occurring in small locally owned stores),^^ These narratives are drawn from the
detailed reporting that exists on 128 of the 289 episodes, as well as previous
ethnographic fieldwork in Argentina (Auyero 2001, 2003), The socio-economic
context of both scenes is the same: The end of 2001 found the residents of poor
enclaves throughout the country struggling to make ends meet with record-high
levels of unemployment (21 percent of the economically active population was
unemployed) and shrinking state assistance. Food-assistance and other welfare
programs (notably, unemployment subsidies) had been steadily declining since
the Congress passed the Law 25,453 publicly known as "Deficit Cero" (Zero
Deficit Law) in July 2001,

The Hipermercado
In December 2001, Liliana Vargas, Angela Caceres, and Roxana Frias were living
in the city of Tucuman, They and their husbands were unemployed, and according
to them had "several kids to feed," Together with another 150 women, they
1354 • Social Forces Volume 85, Number 3 • March 2007

walked to a municipal center where, "we were told the government was giving
out food." When they showed up, however, nobody was there to help them. They
then went to a chain supermarket, Gomez Pardo, were they understood food was
being distributed. As soon as they arrived, police fired rubber bullets and tear gas
at the crowd. "The government lied to us," they said afterward, "we came to ask
for food and they held us back as if we were criminals."
The women's story contains key elements present in our first ideal type:
the lootings that took place in big, chain supermarkets. Poor people initially
gathered in front of these supermarkets because rumors concerning the prompt
distribution of food were running rampant through the neighborhood. Groups of
people, composed mostly of women and children, would arrive and peacefully
assemble at their doors. Once the group had gathered, one of three scenarios
was likely. In one scenario, they were summarily repelled with rubber and/or real
bullets and tear gas by the police or National Guard, as was the case with Liliana
and her friends. In these episodes, no actual looting was allowed to take place. In
another, the police or National Guard were present, but they did not fire their guns
and negotiations ensued. The store manager or owner promised the delivery of
food boxes to avert the looting and members of the crowds negotiated the exact
amount of boxes. If the negotiations were successful, the crowd would dissolve
and again no looting ensued. In many cases these negotiations were supervised
by the police. Negotiations sometimes involved public authorities who, especially
towards the beginning and the end of the week of looting (mainly on Dec. 14
and Dec. 21), showed up with food boxes from state-supported programs to
prevent the looting as in the case of the supermarket Azul (one of the largest
chains in Rosario). In this particular instance, the Secretary of Welfare turned
up in the middle of police using repressive actions and promised the immediate
distribution of food products - "things soon calmed down," a reporter from El
Giudadano wrote on Dec. 15.
In a third possible scenario, negotiations were not altogether successful
even when the police was present. The case of a looting in Chubut illustrates
the "broken-negotiations" dimension of the looting episodes (Tilly 2003). In
this episode, the crowd's delegates accepted the promise of distribution of
merchandise by the manager of the largest supermarket in the city (Comodoro
Rivadavia), but when the content of the boxes was found wanting, the crowds
- vastly outnumbering the police - broke into the premises and took away the
merchandise. Negotiations also went awry for more minor contingencies. A
youngster would through a rock at the window of a supermarket, for example,
or a small group within a much larger crowd would not accept any sort of deal,
triggering mayhem. The typical looting scene in big supermarkets included what
many newspapers called a "looting vanguard," composed mostly of youngsters
and minors who, being used to police harassment, took advantage of this
unique opportunity (having the police outnumbered) to carry out a face-to-face
confrontation with the forces of order.

The Local Market

In December of 2001, Josefa was living in a small shack located in a poor
neighborhood of Moreno. On the 18th, she received a small flyer inviting her to
 Dynamics of Collective Violence • 1355

"bust" a group of markets. The next day, she showed up on tinne in front of Kin
and soon 200 people were gathered in front of this small market claiming food.
She recalls seeing a police car leaving the scene and a man who worked at the
local municipality talking on his cell phone. Soon, a truck loaded with a grupo de
pesados (or group of thugs), known in the neighborhood as "Los Gurkas," arrived
at the scene. "They broke the doors and called us in," Josefa remembers. "A few
days later, I met one of them, and he told me that people from the Peronist Party
paid 100 pesos for the job." Far from Josefa, in another poor enclave in Buenos
Aires, residents of the barrio Baires (located in the municipality of Tigre) seemed
to have received similar news about the imminent looting through their children:
"When my son arrived from school, he told me that a man from the local Unidad
Basica (Peronist Party grassroots office) came to inform the teachers about the
sites of the lootings. The teacher told my son that she was going to go. And we
went to see if we could get something." (Clarin 5/19/02)
This second ideal-type looting scene - one that occurs at a small, locally
owned food market - was noticeably less pregnant with alternative scenarios
than the hipermercado episodes. In these looting episodes, public authorities
were almost never present, and store owners were less prepared to negotiate
with the crowd to more peacefully distribute food. After the groups had gathered
in front of neighborhood supermarkets or grocery stores, owners occasionally
distributed food to avert looting; in other cases, owners barricaded themselves
in their stores and shot at the crowds. These occasions proved to be fatal for
looters: three persons were killed and dozens were injured by store owners. In
most cases, however, nothing other than police repression (which, as we noted
before, tended to be absent in the majority of these cases), prevented the crowds
from breaking into these stores and seizing whatever they could (rumors about
impending police action were usually running rampant at the site so looting
always proceeded at a fast pace). In most episodes when police arrived at all, it
was after the premises were completely sacked.
What seems to be common to most of these episodes is that the crowd
was, in fact, composed of small groups that would arrive together at the looting
site - groups with some level of internal coordination. In 87 of the 128 cases
in which detailed reporting exists, party brokers or other organizers (union
leaders, grassroots activists, militants from organizations of the unemployed)
were present among the claimants or the looters; in many cases they were seen
directing the crowds to and from their targets (Young 2002). The presence of
party brokers and other organizers among many of the looting crowds points to
the existence of previously connected actors giving plausibility to the arguments
made about the existing linkages among participants in joint action, destructive
or otherwise. As Aveni (1973) would say, looters are a "not-so-lonely crowd."
(see also, McPhail and Wohlstein 1983) Recent research on contentious politics
(McAdam et al. 2001; Diani and McAdam 2003) and on collective violence (Tilly
2003) highlights precisely this aspect of collective action episodes: "In practice,"
writes Tilly (2003:32), "constituents' units of claim-making actors often consist
not of living, breathing whole individuals but of groups, organizations, bundles of
social relations, and social sites such as occupations and neighborhoods." In the
1356 • SocialForces Volume85,Number3 • March2007

reminder of this paper we use a subset of these 128 small-marl<et episodes - the
67 cases in which party brokers were reported to be present among the looters
- to delve further into their roles in the dynamics of the food riot.

The Strategic Role of Party Brokers

Far from being an outside organizer, that "mythic culprit blamed so regularly when
'innocent masses' rebel all over the world," (Roy 1984:133) previous research has
demonstrated that political party brokers are deeply embedded in the everyday
life of the poor (Auyero 2001; Brusco et al 2004). In poor and working-class
neighborhoods, shantytowns, and squatter settlements throughout Argentina,
the poor and the unemployed solve the pressing problems of everyday life
(access to food and medicine, for example) through patronage networks that rely
on brokers of the Peronist party (locally known as punteros) as key actors (see also
Levitsky 2003; Torres 2003). Depending to a great extent on the (not always legal,
not always overt) support of the local, provincial and national administrations,
these problem-solving networks work as webs of resource-distribution and of
protection against the risks of everyday life. Punteros provide food in state-funded
soup-kitchens, broker access to state subsidies for the unemployed or to public
hospitals, distribute food and/or food vouchers to mothers, children and elderly,
and occasionally give out toys (manufactured by workfare recipients) to parents
who cannot afford any. As Goldberg (2003, p. 3) writes: "The main source for all
these most basic necessities [food, clothes, and medicine] among impoverished
Argentines is the Peronist neighborhood broker, or puntero." (For a thorough
description of brokers' practices, see Auyero 2001; Levitsky 2003.) Other basic
needs aside, the procurement of food is, according to our own ethnographic
observations and those of other analysts, (Torres 2002; Levitsky 2003; Goldberg
2003) the main task of brokers of the Peronist party.
Brokers channel goods and services from their political patrons to their clients
and political support (in the form of attendance at rallies, participation in party
activities, and sometimes votes) from their clients to their patrons. Punteros
fulfill two main functions in Argentine society: gatekeeping on the one hand and
information-hoarding on the other. Being members of the governing Peronist
party, they have the personal connections that enable them to gain access to
knowledge about resource-distribution. They enjoy what network analysts call
"positional centrality." (Knoke 1990) Neighbors in poor enclaves are usually
unaware of the always changing location of food distribution centers, and the
dates and times when distribution is carried out. Furthermore, they are generally
uninformed of the always changing procedures to obtain the food. Brokers know
the whens, hows and wheres of food allocation.
Most of the literature on patronage networks (wrongly) analyzes them as
antagonistic to contentious collective action. However, some classic works (notably,
Scott and Kerkvliet 1977; Scott 1977) have shown that an interruption in the flow of
vital resources that routinely circulate within them providing for the survival of large
numbers of people can actually trigger collective, often violent, action.
What were the concrete steps and actual mechanisms by which patronage
politics turned into collective violence - clients becoming looters and brokers
 Dynamics of Collective Violence

becoming agitators - during December 2001? The answer to this seemingly

simple question is rather complex. The empirical evidence available thus far tells
us that: 1.) patronage networks feed thousands of poor people on a daily basis;
2.) many of the state-funded social assistance programs that fuel the operation
of these networks were severely curtailed during the second half of 2001; 3.)
brokers are central actors in these networks; 4.) brokers were indeed observed
among many looting crowds; 5.) brokers were present mostly in unprotected
small markets; and 6.) brokers seem to have recruited people to gather in front
of supermarkets spreading news about immediate food distribution with g) some
sort of approval of their patrons (mayors or council members).
Although we can only speculate as to what actually happened on the eve
of the looting week, we have some firm empirical ground to highlight three
mechanisms that, present in other forms of contentious politics (McAdam et al
2001; Tilly 2003), seem to have been central in more than half of the episodes
in which detailed reporting is available: 1.) political brokerage (by which party
brokers connected potential looters among themselves and with their targets);
2.) signaling spirals (by which brokers communicated the location of targets, the
presence or absence of police and thus the feasibility of risky practices), and 3.)
certification (by which public officials, in this case local council members and city
mayors, implicitly or explicitly, authorized brokers to proceed with the lootings).
During the week-long looting brokers of the Peronist party were quite active,
as usual, in search of food for their clients. During this violent week, however,
brokers sought out food neither in the usual places or by the usual means. Given
that the traditional sources of food for their clients (state-funded programs at
municipal, provincial and federal levels) were not responding, brokers turned their
attention to local supermarkets. Demanding food from supermarkets was hardly
an original joint practice. Far from creating a new form of collective violence,
brokers were drawing upon past looting experiences (massive lootings in 1989
for example) and more recent collective claims for food in big supermarkets
that had organizations of the unemployed (locally known as piqueteros) as their
central actors. During the week-long looting, brokers' actions betrayed their
awareness of the presence of police in big chain supermarkets and of its absence
in neighborhood stores. Again, they were drawing upon local knowledge: small
markets cannot afford special state protection. Either following the explicit orders
of their patrons or (what is more likely given the decentralized character of the
Peronist Party)'^ acting on their own but with the tacit approval of their leaders,
they searched for food for and with their clients by means other than they usual
ones: they looted.
Even though flyers inviting people to attack supermarkets did indeed circulate,
and although there are reports that say that brokers took prospective looters to
the sites, the process seems to be less straightforward. Through their networks,
brokers publicized information (or simply rumours) concerning the upcoming
distribution of food in local supermarkets. As a prominent Peronist broker from
the district of Moreno matter-of-factly told us: "We (the members of the party)
knew about the lootings beforehand, around 1 a.m. (lootings began by noon)
we knew that there was going to be a looting ("sabiamos que se iba a saquear").
1358 • Social Forces Volume 85, Number 3 • March 2007

Municipal officials told us so. We passed the information along..." Given their
own reputation as food providers, residents of poverty enclaves acted on this
information broadcasted by brokers and began gathering in front of these stores.
With hundreds and sometimes thousands of desperate people believing in
the imminent giving out of food gathered in front of unguarded stores, minor
contingencies then determined the unfolding or not of collective violence.
To fully understand and explain the transition from clientelist politics to
collective violence, further qualitative research is needed. In particular, we need
to investigate more closely how the brokers' create a window of opportunity for
looting, as well as on the processes through which poor people transition from
"food seekers" to "looters." In the meantime, however, we can confidently make
the argument that, at least in the majority of the lootings in which good, detailed
reporting exists, the rootedness of collective violence in daily political life can be
better grasped by paying due attention to patronage networks and the specific
actions of party brokers.

Conclusion
Food riots are diverse, complex and dynamic forms of collective behavior. In
this article, we presented interrelated quantitative and qualitative analyses of a
particular series of food riots to scrutinize some dimensions of this still under-
explored but all too present form of collective violence. We found that far from
being a monolithic wave, the lootings were locally diverse in terms of their
geographic distribution, number of participants, type of market sacked, extent
of state repressive action, and whether political party brokers were involved. We
then focused more closely on the inner dynamics of the looting episodes, paying
particular attention to the interactions between police action, type of market
and the presence of political party brokers. The actions of political brokers and
police officers during the lootings offered an unparalleled window to examine
a set of problems vitally important to the social scientific study of collective
violence, which has rarely been examined head-on, namely the deep grounding
of ostensibly spontaneous looting in routine forms of political activity, and the
intimate interaction between damage-making collective action and official
repressive action.
After recreating two ideal-type looting scenarios, one emphasizing the role
of the police in large, chain supermarkets, the other concentrating on the role
of brokers in looting smaller, locally owned shops, this research analyzed the
lootings that reported brokers at the site, examining the working of party networks
in every day life and the possible mechanisms that went into operation during
the week of Dec. 14-21, 2001. Existing scholarship insists on the rootedness
of collective violence in routine social relations (Blok 2001; Rule 1988), on the
multifarious ways in which violent contention takes place embedded, and often
hidden, in the mundane structures of everyday life and routine politics (Roy
1994). As Tilly (1992:6) writes: "Contentiousgatheringsobviously bear a coherent
relationship to the social organization and routine politics of their settings. But
what relationship? That is the problem." This research addresses precisely
 Dynamics of Collective Violence •1359

this problem by focusing on patronage networks as key connectors between

routine and non-routine collective action, providing solid empirical evidence for
a generally unrecognized and understudied way in which routine politics carries
over into the processes of collective violence.
Attention paid to brokers and their networks, and to the concrete mechanisms
by which networks change their function, allows us to (re)focus on /nrez-actions
between actors, blurring the lines separating "non-routine" and "routine"
collective action (a dichotomy that remains at the core of breakdown theories
of collective action [Useem 1998; Snow et al. 1998]). Eurthermore, close-up
examination (both qualitative and quantitative) of the state and workings of
patronage problem-solving networks, and particularly of the role brokers play
within them during "normal" political life and of the role they might have played
during the lootings, is needed to fully understand and explain the embeddedness
of collective violence in normal social life. Further research is also needed on
the relationship between strength/weakness of clientelist networks and levels
of popular participation in food riots.

Notes
1. There is also a considerable amount of work done on repression as a
dependent variable. For recent reviews, see Earl (2003) and Earl et al. (2003).
See also Wisler and Giugni (1999).

2. Food lootings took place in every Latin American country (with the exception of
Chileand Paraguay) during the past 15years. Either in the context of subsistence
crises brought by natural disasters (droughts, flooding, earthquakes) or in the
context of suddenly imposed grievances due to politico-economic crises or in
the margins of collective protests against neo-liberal policies, Latin American
countries have witnessed hundreds and sometimes thousands of desperate
people attacking food markets and taking away the merchandise.

3. For an insightful and systematic examination of the role of politicians and

state authorities in ethnic riots in India, see Wilkinson (2004).

4. President De la Rua governed from May 1999 until December 2001 when
he resigned in the midst of the unprecedented politico-economic crisis and
mass protests. A quick succession of two different presidents (all belonging
to the Peronist Party) ended when Duhalde, a former governor of Buenos
Aires, was elected by the Parliament to become interim president.

5. For a recent comprehensive review on the problems and possible solutions

involving the use of newspapers see Earl et al. 2004.

6. Because reliability tests are usually standard in these situations, after

coding was complete we randomly selected a 50 percent sub-sample of
the newspaper articles and re-coded them. In every case the original coding
was obtained. Because we are coding "hard news items" no subjective
interpretations need to be made (for example, the news article states that
police were there or it does not). This is not to say that there is no reporting
1360 • Social Forces Volume 85, Number 3 • March 2007

bias in the subjective interpretations of the journalists themselves, just not in

our coding of what they wrote.

7. There are potentially instances where an article fails to report broker (or police)
presence. If this bias exists in our dataset, however, it is in a methodologically
conservative direction - it would lead us to underestimate the coordination
of their activities. If anything then, accounting for this bias would serve to
reinforce our central arguments below.

8. Details of the ethnographic research can be found in Auyero (2001, 2003).

9. Interestingly, in the provinces in which data is available, the overwhelming

majority of those arrested had no penal records - confirming what classic
studies on rioting (Caplan 1970; Capian and Paige 1968; Moinat et al. 1972)
assert - looters are not the tiny criminal minority in poverty enclaves.

10. Stark et al (1974) and Varhney (2002) make the same point for the case of
race and ethnic riots.

11. The Peronist party was in power from 1989 to 1999 (under President
Menem) when it eventually lost to the Alianza (a coalition of center and
center-left parties). Although not in charge of the national administration
in 2001, the Peronist party held the majority of state governorships and
of parliament members. A huge party with vast membership and activist
base, and extensive local-level organization, the Peronist Party is deeply
rooted in the working- and lower-class society of Argentina (though not
exclusively) through machine-style politics. Levitsky's (2003) book is the
best English language source to understand the recent history and the
workings of the Peronist Party.

12. On the analytical accentuation and empirical synthesis involved in the

construction of ideal-types, see Weber (1949, p. 90).

13. On the decentralized character of the Peronist Party, see Levitsky's (2003)
insightful analysis.

References
Amin, Shahid. 1995. Event, Metaphor, Memory. Chauri-Chaura 1922-1992.
California University Press.

Amato, Alberto. 2002. "La Trama Polftica de los Saqueos." Glarfn Digital. Dec. 19.

Armstrong, Gary. 1998. Footbail Hooligans. Knowing the Score. Berg.

Arias, Desmond. 2003. "The Infrastructure of Criminal Governance: Illegal

Networks and Public Order in Rio de Janeiro." Dallas, Texas: Paper Presented
at the 2003 Meeting of the Latin American Studies, March 27-29.

Auyero, Javier 2003. Contentious Lives. Duke University Press.

 Dynamics of Collective Violence •1361

. 2001. Poor People's Politics. Duke University Press.

Aveni, Adrian. 1977. "The Not-So-Lonely Crowd: Friendship Groups in Collective

Behavior." Sociometry 40(1 ):96-99.

Bessinger, Mark. 1998. "National Violence and the State." Comparative Politics
30:401-22.

Blok, Anton. 2001. Honor and Violence. Blackwell.

Bohstedt, John. 1983. Riots and Community Politics in England and Wales 1790-
1810. Cambridge University Press.

Bonasso, Miguel. 2002. El Palacioyla Calle. Planeta.

Bourgois, Philippe. 2003. In Search of Respect Selling Crack in El Barrio.

Cambridge University Press.

Bouton, Cynthia. 1993. The Flour War Gender, Class, and Community in Late
Ancien Regime French Society. Pennsylvania State University Press.

Brass, Paul. 1997. Theft of an Idol. Princeton University Press.

Braun, Herbert. 1980. The Assassination of Gaitan. Public Life and Urban Violence
in Colombia. University of Wisconsin Press.

Brusco, Valeria, Marcelo Nazareno and Susan Stokes. 2004. "Vote Buying in
Argentina." Latin American Research Review 39(2):66-88.

Camarasa, Jorge. 2002. Di'as de Furia. Sudamericana.

Cafassi, Emilio. 2002. Olla a Presion. Cacerolazos, Piquetes y Asambleas, sobre

Fuego Argentino. Libros del Rojas.

Caplan, Nathan. 1970. "The New Ghetto Man: A Review of Recent Empirical
Studies." Journal of Social Issues 26:59-73.

Caplan, Nathan, and Jeffrey Paige. 1968. "Survey of Detroit and Newark Riot
Participants." Pp. 127-37. Report of the National Advisor Commission on Civil
Disorders. Bantam Books.

Colectivo Situaciones. 2002. 19y20Apuntespara el Nuevo Protagonismo Social.

Ediciones de Mano en Mano.
Danzger, M.H. 1975. "Validating Conflict Data." American Sociological
40:570-84.

Davenport, Christian. Editor. 2000. Paths to State Repression. Rowman &

Littlefield Publishers.

Davis, Natalie. 1973. "The Rites of Violence: Religious Riot in Sixteenth-Century

France." Past and Present 59:51-91.
1362 • Social Forces Volume 85, Number 3 • March 2007

Das, Veena. 1990. Editor. Mirrors of Violence. Communities, Riots, and Survivors
in South Asia. Oxford University Press.

Delia Porta, Donatella. 1995. Political Movements, Political Violence, and the
State. Cambridge University Press.

. 1996. "Social Movements and the State: Thoughts of the Policing

of Protest," Pp. 62-92. Comparative Perspectives on Social Movements.
Doug McAdam, John D. McCarthy and Mayer N. Zaid, editors. Cambridge
University Press.

Diani, Mario, and Doug McAdam. Editors. 2003. Social Movements and Networks.
Relational Approaches to Collective Action. Oxford University Press.

Dynes, R., and E.L. Ouarantelli. 1968. "What Looting in Civil Disorder Really
Means." Transactions 5:9-14.

Earl, Jennifer, Sarah A. Soule and John McCarthy 2003. "Protest under Fire?
Explaining the Policing of Protest." American Journal of Sociology 68:58^-606.

Earl, Jennifer, Andrew Martin, John D. McCarthy and Sarah A. Soule. 2004. "The
Use of Newspaper Data in the Study of Collective Act\o'r\." Annual Review of
Sociology 30:65-80.

Elison, Graham, and Greg Martin. 2000. "Policing, Collective Action, and Social
Movement Theory: the Case of the Northern Ireland Civil Rights Campaign."
British Journal of Sociology. 51(4):681-99.

Erickson, Bonnie. 1996. "The Structure of Ignorance." Charleston, S.C: Keynote

Address, Sunbelt XVI: International Sunbelt Social Network Conference, Feb. 22.

Feagin, Joe, and Harlan Hahn. 1973. Ghetto Revolts. The Politics of Violence in
American Cities. Macmillan.

Franzosi, Roberto. 1987. "The Press as a Source of Sociohistorical Data." Historical

Methods 20:5-^6.

Fradkin, Raul. 2002. Cosecharas tu Siembra. Prometeo Libros.

Gamson, William A. [1975] 1990. The Strategy of Social Protest. Dorsey.

Giarraca, Norma. Editor. 2001. La Protesta Social en la Argentina. Alianza.

Goldberg, Jonathan. 2003. "Campaign Conscripts. How to fill a stadium with

Argentina's poor (and other ways to win the presidency)." The American
Prospect (April) On Line Edition. Accessed Aug. 23, 2003 at: http://www.
prospect.org.

Goldstein, Donna. 2003. Laughter Out of Place. Race, Class, and Sexuality in a Rio
Shantytown. University of California Press.
 Dynamics of Collective Violence •1363

Gould, Roger. 'iddS. Insurgent Identities. Class, Community, and Protest from 1848
to the Commune. Chicago University Press.

Greenberg, Cheryl. 1992. "The Politics of Disorder: Reexamining Harlem's Riots

of 1935 and 1943." Journal of Urban History 18:395-441.

Gurr, Ted. 1970. Why Men Rebel. Princeton University Press.

Hibbs, Douglas A. 1973. Mass Political Violence. John Wiley & Sons.

Hirsch, Eric L. 1990. "Sacrifice for the Cause: Group Processes, Recruitment,
and Commitment in a Student Social Movement." American Sociological
Review 55:243-54.

Kakar, Sudhir. 1996. The Colors of Violence. Cultural Identities, Religion, and
Conflict. University of Chicago Press.

Killian, Lewis. 1980. "Theory of Collective Behaviour: The Mainstream Revisited."

Pp. 275-89. Sociological Theory and Research: A Critical Appraisal. Hubert
Blalock, editor. Free Press.

Knoke, David. 1990. Political Networks. Cambridge University Press.

Kohan, Anibal. 2002. A las Calles! Una Historia de los Movimientos Piqueteros y
Caceroleros de los '90 al2002. Ediciones Colihue.

Kollman, Raul. 2001. "Opinion." Paginal2\/\/eb. Dec. 20.

Koopmans, Ruud, and Dieter Rucht. 1999. "Protest Event Analysis - Where to
Now?" Mobilization 4:123-30.

. 2002. "Protest Event Analysis." Pp. 231-59. Methods of Social Movement

Research. Bert Klandermans and Susan Staggenborg, editors. University of
Minnesota Press.

Kornblum, William. 1974. Blue Collar Community. University of Chicago Press.

Kriesi, Hanspeter, R. Koopmans, J.W Dyvendak and M. Giugni. 1995. New Social
Movements in Western Europe. University of Minnesota Press.

Levine, Felice, and Katherine Rosich. n/d. Social Causes of Violence: Crafting a
Science Agenda. Accessed Jan. 30, 2003 at: www.asanet.org.

Levitsky, Steve. 2003. Transforming Labor-Based Parties in Latin America: Argentine

Peronism in Comparative Perspective. Cambridge University Press.

Lewkowics, Ignacio. 2002. Sucesos Argentinos. Paidos.

Markoff, John. 1996. The Abolition of Feudalism. Peasants, Lords, and Legislators
in the Erench Revolution. Pennsylvania State University Press.
1364 • SocialForces Volume85,Number3 • March2007

Marx, Gary. 1974. "Thoughts on a Neglected Category of Social Movement

Participant: The Agent Provocateur and the Informant." American Journal of
Sociology 80{2):A02-A2.

Martinez, Tomas. 2002. Episodios Argentinos. Aguilar.

McAdam, Doug. 1982. Poiiticai Process and the Development of Black Insurgency
1930-1970. University of Chicago Press.

. 1988. Freedom Summer. Oxford University Press.

McAdam, Doug, Sidney Tarrow and Charles Tilly. 2001. Dynamics of Contention.
Cambridge University Press.

McPhail, Clark. 1971. "Civil Disorder Participation: A Critical Examination of

Recent Research." American Sociological Review 36:1058-73.

. 1991. The Myth of the Madding Crowd. Aldine de Gruyter.

. 1992. Acting Together: The Organization of Crowds. Aldine de Gruyter.

McPhail, Clark, and Ronald Wohlstein. 1983. "Individual and Collective

Behaviors within Gatherings, Demonstrations, and Riots." Annual Review
of Sociology 9:579-600.

Moinat, Sheryl, Walter Raine, Stephen Burbeck and Keith Davison. 1972., "Black
Ghetto Residents as Rioters." Journal of Social Issues 28:45-62.

Mueller, Carol. 1997. "International Press Coverage of Eastern German Protest

Events." American Sociologicai Review 62:820-32.

Myers, D. 1997. "Racial Rioting in the 1960s." American Sociological

Review 62:9A-^^2.

Myers, D., and Beth Schaefer Caniglia. 2004. "All the Rioting That's Fit to Print:
Selection Effects in National Newspaper Coverage of Civil Disorders, 1968-
1969." American Sociological Review 69:519-43.

Olzak, S. 1989. "Analysis of Events in Studies of Collective Action."/^/7n(ya//?ei//eiv

o/Soc/b/05'/15:119-41.

. 1992. The Dynamics of Ethnic Competition and Conflict. Stanford

University Press.

Piven, Frances Fox, and Richard A. Cloward. 1977. Poor People's Movements:
Whey They Succeed, How They Eail. Vintage.

Roldan, Mary. 2002. Blood and Fire. La Violencia in Antioquia, Colombia, 1946-
1953. Duke University Press.
 Dynamics of Collective Violence •1365

Rosenfeld, Michael. 1997. "Celebration, Politics, Selective Looting and

Riots: A Micro Level Study of the Bulls Riot of 1992 in Chicago." Social
Problems 44(4): 483-502.

Roy, Beth. 1994. Some Trouble with Cows. California University Press.

Rude, George. 1964. The Crowd in History. John Wiley & Sons.

Rule, James. 1988. Theories ofGivil Violence. California University Press.

Sain, Marcelo, 2004. Politica, Policia y Delito. La red bonaerense. Capital IntelectuaL

Salert, Barbara, and John Sprague. 1980. The Dynamics of Riots. Inter-University
Consortium for Political and Social Research.

Sanchez, Gonzalo, and Donny Meertens 2001. Bandits, Peasants, and Politics:
The Case of "La Violencia" in Colombia. Texas University Press.

Schneider, Jane, and Peter Schneider. 2003. Reversible Destiny. Mafia, Antimafia,
and the Struggle for Palermo. California University Press.

Scott, James, and Benedict J. Kerkvliet. 1977. "How Traditional Rural Patrons
Lose Legitimacy: A Theory with Special Reference to Southeast Asia." Pp.
439-58. Eriends, Eollowers, and Eactions: A Reader in Political Glientelism.
Laura Guasti, Carl Lande, James Scott and Steffen Schmidt, editors. California
University Press.

Scott, James. 1977. "Patronage or Exploitation?" Pp. 67-94. Patrons and

Clients in Mediterranean Societies. Ernest Gellner and John Waterbury,
editors. Duckworth.

Shaheed, Farida. 1990. "The Pathan-Muhajir Conflicts, 1985-6: A National

Perspective." Pp. 194-214. Mirrors of Violence. Communities, Riots and
Survivors in South Asia. Veena Das, editor Oxford University Press.

Snow, David, Daniel M. Cress, Liam Downey and Andrew W. Jones. 1998.
"Disrupting the 'Quotidian': Reconceptualizing the Relationship Between
Breakdown and the Emergence of Collective Action." Mobilization 3(1): 1-22.

Stark, Margaret Abudu, Walter Raine, Stephen Burbeck and Keith Davison.
1974. "Some Empirical Patterns in a Riot Process." American Sociological
Review 39:865-76.

Svampa, Maristella, and Sebastian Pereyra. 2003. Entre la Ruta y el Barrio. La

Experienda de las Organizaciones Piqueteras. Biblos.

Tarrow, Sidney. 1998. Power in Movement. Social Movements and Contentious

Politics. Cambridge University Press.
1366 • SocialForces Volume85,Number3 • March2007

Taylor, Lynne. 1996. "Food Riots Revisited." Jouma/o/Soc/a/W/sto/y(Winter):483-96.

Thompson, E. R 1994. Customs in Common. The New Press.

Tilly, Charles, Louise Tilly and Richard Tilly 1975. The Rebellious Century, 1830-
1930. Harvard University Rress.

Tilly, Charles. 1978. From Mobiiization to Revoiution. Addison Wesley

. 1986. The Contentious French. Harvard University Press.

. 1992. "How to Detect, Describe, and Explain Repertoires of Contention."

Center for the Study of Social Change, New School for Social Research. The
Working Paper Series 150:6.

_. 2000. "Spaces of Contention." Unpublished Manuscript. Columbia

University.

. 2003. The Politics ofCoiiective Vioience. Harvard University Press.

Torres, Pablo. 2002. Votos, Chapasy Fideos. De la Campana.

Useem, Bert. 1985. Disorganization and the New Mexico Prison Riot of 1980.
American Socioiogicai Review 50:677-88.

. 1998. Breakdown Theories of Collective Action. Annual Review of

24:215-38.

Varshney, Ashutosh. 2002. Ethnic Confiictand Civic Life. Hindus and f\Auslims in
india. Yale University Press.

Villalon, Roberta. 2002. "Piquetes, Cacerolazos y Asambleas Vecinales: Social

Protests in Argentina, 1993-2002." MA Thesis. University of Texas at Austin.

Volkov, Vadim. 2002. Vioient Entrepreneurs. The Use of Force in the Mai<ing of
Russian Capitaiism. Cornell University Press.

Walton, John. 1989. "Debt, Protest, and the State in Latin America." Pp. 299-
328. Power and Popular Protest Latin American Social Movements. Susan
Eckstein, editor. California University Press.

Walton, John, and Charles Ragin. 1990. "Global and National Sources of Political
Protest: Third World Responses to the Debt Crisis." American Socioiogicai
Review 55:876-90.

Walton, John, and David Seddon. 1994. Free f\4arkets and Food Riots. The Poiitics
of Global Adjustment. Blackwell.

Walton, John, and Jon Shefner. 1994. "Latin America: Popular Protest and the
State." Pp. 97-134. Free i\/larkets and Food Riots. John Walton and David
Seddon, editors. Blackwell.
 Dynamics of Collective Violence '1367

Weber, Max. 1949. The Methodology of the Sociai Sciences. Free Press.

Wilkinson, Steven. 2004. Votes and Vioience. Cambridge University Press.

Wohlenberg, Earnest H. 1982. "The 'Geography of Civility' Revisted: New York

Blackout Looting, 1977." Economic Geography 58:29-44.

Young, Gerardo. 2002. "La Trama Polftica de los Saqueos," Clarin Digitai. Dec. 19.

Zhao, Dingxin. 2001. The Power of Tiananmen. State-Society Reiation and the
1989 Beijing Student Movement. Chicago University Press.