PreprintPDF Available

The Symbolic Logic of Pogrom Violence: Evidence from Kristallnacht

Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

What explains why pogroms result in violence in some places, but not in others? The literature on similar forms of collective violence advances four separate causal explanations: (1) pogrom violence occurs where its targets are most visible; (2) violence occurs where pogrom organizers face the greatest political competition; (3) violence occurs where hatred of the targeted group is most pronounced; and (4) violence occurs where the surrounding society is least capable of resisting it. I describe the causal logic behind each of these arguments and present testable hypotheses associated with each. To evaluate these hypotheses, I introduce a new geocoded dataset of Jewish synagogues and prayer houses in Germany, based on information gathered by a Jewish memorialization project, to examine municipal-level variation in anti-Semitic violence during the November 1938 Kristallnacht ("Night of Broken Glass") pogrom. Using multivari-ate regression analysis, I find that pogrom violence is most likely and relatively more severe where the targeted population is most visible, and that a higher concentration of the pogrom organizers' political opponents increases the absolute severity of group-selective violence. These findings provide evidence for the symbolic logic of pogrom violence. I conclude by discussing theoretical, empirical, and methodological implications for the study of pogroms and related forms of political violence.
Content may be subject to copyright.
The Symbolic Logic of Pogrom Violence:
Evidence from Kristallnacht
Daniel Solomon
August 22, 2019
What explains why pogroms result in violence in some places, but not in others? The lit-
erature on similar forms of collective violence advances four separate causal explanations: (1)
pogrom violence occurs where its targets are most visible; (2) violence occurs where pogrom or-
ganizers face the greatest political competition; (3) violence occurs where hatred of the targeted
group is most pronounced; and (4) violence occurs where the surrounding society is least capable
of resisting it. I describe the causal logic behind each of these arguments and present testable
hypotheses associated with each. To evaluate these hypotheses, I introduce a new geocoded
dataset of Jewish synagogues and prayer houses in Germany, based on information gathered by
a Jewish memorialization project, to examine municipal-level variation in anti-Semitic violence
during the November 1938 Kristallnacht (“Night of Broken Glass”) pogrom. Using multivari-
ate regression analysis, I find that pogrom violence is most likely and relatively more severe
where the targeted population is most visible, and that a higher concentration of the pogrom
organizers’ political opponents increases the absolute severity of group-selective violence. These
findings provide evidence for the symbolic logic of pogrom violence. I conclude by discussing
theoretical, empirical, and methodological implications for the study of pogroms and related
forms of political violence.
1 Introduction
On November 9th, 1938, senior officials of Germany’s Nazi regime set off a brief wave of acute
violence against the country’s Jewish population. In two days, Nazi paramilitaries, youth militias,
civilian members of the Nazi Party, and state security forces laid to waste the last remnants of
Germany’s robust Jewish society. The pogrom, commonly called Kristallnacht (the “Night of
Department of Government, Georgetown University, I am thankful for comments I
received on earlier drafts of this paper from Michael Bailey, Laia Balcells, Matthew Kocher, Jeffrey Kopstein, Yonatan
Lupu, Harris Mylonas, Livia Schubiger, and Robert Shapiro, and from participants in the 2019 DC Comparative
Politics Workshop, students in the Fall 2018 Civil War and Sub-state Violence seminar at Georgetown University,
students in the Spring 2019 Nationalism and Nation-Building seminar at George Washington University, and attendees
of the 2019 Association for the Study of Nationalities World Convention. I also thank Jon Askonas, Lenore Bell, Marc
Grellert, Lise Howard, David Laitin, Aliza Luft, Hans Noel, Lucy Solomon, and Yuhki Tajima for their suggestions,
conversation, and research guidance. All errors are my own.
Broken Glass”), was not the first episode of Nazi mass violence against German Jews and their
communal institutions—nor, we now know, the last. The scale of violence during Kristallnacht,
however, was a dramatic escalation from the previous five years of public abuse, harassment, and
government-sanctioned property seizures that had characterized the German Jewish experience
since Adolf Hitler and the Nazi Party’s rise to power in 1933.
Anti-Semitic violence during the Kristallnacht pogrom ranged widely, but the episode did
not affect all Jewish communities equally. Party members harassed and attacked their Jewish
neighbors; Nazi paramilitaries defaced Jewish storefronts; members of the Nazi security services
deported tens of thousands of Jews to concentration camps in Germany. In the most memorable
aspect of the pogrom, perpetrators destroyed hundreds of synagogues and prayer houses across
Germany. Approximately one-third of Jewish houses of worship in Germany, however, displayed
no evidence of damage. What explains why perpetrators attacked synagogues in some areas, but
not others? More generally, what explains why pogroms result in varying levels of violence against
targeted groups?
In this paper, I articulate and test competing answers to this question. The social science
literature on episodes of collective violence similar to pogroms is substantial. Political scientists
and sociologists, in particular, have sought explanations for riotous violence at various levels of
analysis since the social unrest of the American postwar period (Spilerman, 1971). New methods
of comparative empirical inquiry and the growing availability of fine-grained data about violent
behavior, however, mean that prevailing explanations for this violence warrant new inquiry. Until
recently, the literature on riots and collective violence has also stopped short of offering systematic
explanations for a paradigmatic case of group-selective violence, the Nazi Holocaust against Jewish
communities throughout Europe (King, 2012; Kopstein and Wittenberg, 2018).
I proceed in five parts. First, I define pogroms as a distinct form of large-scale, group-selective
violence. Variations in violence during these episodes are theoretically puzzling, as there are few
social costs to participation in collective violence and local agents of violence have significant auton-
omy over violent mobilization. Second, I draw on the literatures on political violence, contentious
politics, and inter-group relations to identify four solutions to this puzzle. Third, I outline a re-
search design to test these theories, drawing on a new geocoded dataset of attacks against Jewish
synagogues and prayer houses during the Kristallnacht pogrom. Fourth, I present evidence that
pogrom organizers use violence for symbolic purposes, and that political competition is also a pre-
tense for more widespread attacks against targeted groups. I conclude by discussing this argument’s
implications for the theoretical and empirical study of violence in political science.
2 Defining the Pogrom
A pogrom is a relatively brief episode of multiple, public violent acts by an informal group against
a specific social community. Historians typically associate the term “pogrom” with a bounded wave
of violence against Jewish communities in the Eastern European Pale of Settlement during the late
19th and early 20th century (Klier and Lambroza, 1992). Scholars of other episodes of collective
violence, however, use the term to describe interethnic violence in non-European contexts, such as
India (Brass, 2003; Ghassem-Fachandi, 2012) and Indonesia (Sidel, 2006) as well as violence against
Eastern European Jews during other historical periods. In some contexts, scholars may use other
terms to describe events that share the definitional characteristics of a pogrom, such as race riots
(Ellsworth, 1992), ethnic riots (Horowitz, 2001), and collective or communal violence (Tambiah,
1997; Tajima, 2014; Krause, 2018).
A pogrom episode is relatively brief because there is no unambiguous formula for calculating its
chronological boundaries across different social contexts. The metaphorical language of rapidity—
”intense, sudden” (Horowitz, 2001); “spark upon a bed of combustible material” (Brass, 1996),
“wave” (Kopstein and Wittenberg, 2018)—is common in the literature about pogroms, riots, and
collective violence. These metaphors point to the analytic difficulty of comparing pogroms across
uniform time units. Some pogroms, such as the events of Kristallnacht, take place in a matter of
days, while other instances of similarly-patterned violence may stretch over a longer period of time
(Brass, 1996). The onset of pogrom violence is straightforward—it begins when the violence starts—
but the dynamism of the episode makes their endings more difficult to anticipate. Whether pogrom
violence on one day leads to violence the next depends on the decisions of semi-autonomous agents
involved in mobilizing repeated acts of collective violence, rather than the regimented program of
a military command (Tilly, 2003).
These semi-autonomous agents of pogrom violence are members of an informal group. By
informal, I refer to the loose, unofficial organization of perpetrators involved in pogrom attacks.
This organization typically differs from violence organized by state security forces or non-state rebel
groups, in which the direct perpetrators of violence regularly use force in the course of their sep-
arate professional or social responsibilities. The informal group of pogrom organizers can include
members of military or paramilitary forces, but the command structure of those organizations is
formally unaffiliated with the pogrom network. Instead, pogroms rely on a “network” (Kopstein
and Wittenberg, 2018) or “institutionalized riot system” (Brass, 1997) to activate violence. This
description of the organization of pogroms departs from popular understandings of pogroms as a
form of “civilian-against-civilian” violence. Although pogroms involve civilians, the active agents
of physical violence are more commonly “specialists” (Brass, 1997) whose personal biographies and
collective networks make mobilization possible. Many pogrom organizers are former members of
military or paramilitary groups, despite lacking regular employment as security officials. Formerly
incarcerated members of Polish and Ukrainian nationalist organizations, for example, were central
to the anti-Semitic pogroms in Poland’s eastern borderlands during the summer of 1941 (Lower,
2011). These pogrom organizers were “neighbors” (Gross, 2001)—they often lived in close prox-
imity to their Jewish victims, as opposed to dwelling in separated barracks—but their previous
participation in violent insurgencies complicates their “civilian-ness.”
Pogrom organizers and their local counterparts jointly produce pogrom violence. In this re-
spect, the social process of a pogrom episode bears greater resemblance to non-violent or quasi-
violent social movements than to other instances of mass political violence like civil war or genocide.
The social relationships and resources that allow for collective action by social movements mirror the
constituent dynamics of a pogrom (Scacco, 2008). All pogroms require that organizers coordinate
the public actions of a spatially disparate group of people, similar to the process of coordinating a
public protest or other demonstration. Pogrom organizers distribute information and resources—
transportation, weapons, and other means of collective action—to local networks, which enable
those networks to act in conjunction with others outside their locality for a concerted social pur-
pose. As with the time units of a pogrom episode, the spatial boundaries of these networks vary
across time and space.
Pogrom organizers also provide their local counterparts with a common target, the impetus
for collective action. The descriptive characteristics of the pogrom target differentiate the pogrom
from its broader conceptual cousin, the riot. A riot may serve multiple purposes for its organizers
that do not relate to the ascriptive characteristics of its targets. Consider US Black freedom
organizer Martin Luther King, Jr.’s lyric explanation of Black riots in the United States as the
“language of the unheard”; for Black freedom groups in the United States, violence was part of
the repertoire of public protest rather than a means of group-selective intimidation (Piven and
Cloward, 1977). In 19th-century England, tenant laborers organized the Swing riots to protest the
mechanization of British agriculture. Rioters lit fires and destroyed farming equipment, tactics of
property destruction that, at a glance, resemble those used by pogrom organizers. These riots,
however, were individually rather than group-selective: rioters destroyed the threshing machines of
specific farmers associated with mechanization, rather than viewing the landed class as a general
target of violence (Hobsbawm and Rud´e, 1969). Pogrom violence is always group-selective: an
individual’s membership in the targeted group means that they and their property are more likely
to experience physical harm after the onset of a pogrom than non-members.1In mobilizing the
group-selective pogrom, organizers also deploy rhetorical and visual tropes that cast members of
the group, in toto, as objects of social exclusion.
Unlike the perpetrators of similarly time-bounded episodes of mass violence like massacres or
lynchings, pogrom organizers deploy a diverse “repertoire” of violent techniques against the targeted
group (Wood, 2009; Hoover Green, 2018; Guti´errez-San´ın and Wood, 2017). These include killing,
rape and other forms of sexual violence, the destruction of public symbols or communal sites
associated with the targeted group, the destruction of property and businesses belonging to the
targeted group, and acts of public harassment and maiming. These acts of violence take place in
public view, consistent with the performative logic of a protest or other type of contentious action.
The pogroms literature makes much of the gruesomeness of some of these techniques (Horowitz,
2001), many of which align with Fujii’s concept of “extra-lethal violence”: ”physical acts committed
face-to-face that transgress shared norms and beliefs about appropriate treatment of the living as
1(Steele, 2017) terms this type of violence “collective targeting.” Meanwhile, the literature on collective violence
and contentious politics (Tilly, 2003) refers to the organizers of the violence, rather than targeted groups, as a
“collective.” I prefer “group-selective violence” because it refers unambiguously to the targets of violent mobilization.
well as the dead” (Fujii, 2013, p. 410). Whether extra-lethal violence is unique to pogroms, or
instead a ubiquitous feature of mass-violence episodes, is an outstanding puzzle for the study of
political violence.
As the above discussion indicates, the definitional boundaries of a pogrom episode are fuzzy and
context-specific. Despite common patterns across cases, the specific organizational characteristics
of pogrom organizers, the ascriptive features of the targeted group, and the tactics and means of
coordination that organizers employ emerge from a particular set of social and historical conditions,
and will change accordingly. The context-specificity of a pogrom is a function of the time-bounded
political events that contribute to its onset, in addition to the sociocultural differences between the
societies in which the episode takes place. In Indonesia in late 1998, instances of large-scale pogrom
violence by insurgent Islamist groups aligned with the growing political power of Islamic parties in
the new civilian government. Pogroms in similar locations just two years later took place under
conditions of displacement-related segregation and the new government constraints on the networks
responsible for the pogroms (Sidel, 2006). The particular meso-level structures and processes that
precede the onset of the violence are a critical feature of any causal explanation for the localities
in which a pogrom does or does not take place. We now turn to four meso-level explanations for
this variation.
3 Competing Explanations for Pogrom Violence
Why does violence occur in some places after the onset of a pogrom, but not in others? The
literature on pogroms and similar forms of violence advances four separate causal explanations:
(1) pogrom violence occurs where its targets are most visible; (2) violence occurs where pogrom
organizers face the greatest political competition; (3) violence occurs where hatred of the targeted
group is most pronounced; and (4) violence occurs where the surrounding society is least capable of
resisting it. The social complexity of pogrom episodes means that some combination of each of these
factors contributes to the diffusion of pogrom violence. We cannot understand the phenomenon of
pogrom violence in Nazi Germany, for example, without reference to historical patterns of political
anti-Semitism in Germany, or to the collapse of Weimar civil society under Nazi authoritarianism.
The task of causal inquiry, however, is to identify the common tendencies that differentiate the
sites in which pogrom violence is most pronounced, from those where it is not. In the following
section, I discuss the theoretical basis for each of these explanations and describe their observable
implications in the form of testable hypotheses.
3.1 Symbolic Violence
Like other social movements, pogrom organizers deploy symbols to convey their movement frames
to their constituents, targets, and external observers (Snow and Benford, 1992). The most obvious
movement symbols are visual insignia or imagery, such as the Nazi swastika or images of Nazi rallies,
or rhetorical, as in official speeches or the songs, mottos, and chants that movement organizers use to
rally their members. Symbols are intersubjective, in that a society’s shared understanding of—and
disagreement about—symbolic meanings define their social purpose. When used by movements,
these public symbols demonstrate the social authority of both the organizers and the movement as a
whole (Turner and Surace, 1956). As Tambiah writes in his description of Indian religious festivals
as sites of collective violence, “[t]he more the marchers, the louder their slogans, the more macho
their getup, the more ’powerful’ the politician” (Tambiah, 1997, p. 240). A public concentration of
this iconography amounts to a “show of force,” which generates attention to and acknowledgement
of the movement’s ability to occupy the proverbial public square.
The violence of the pogrom episode is itself a symbolic practice (Bourdieu, 1980, 1993). Because
pogroms take place in public, pogrom violence functions as a form of “violent display” (Fujii, 2017).
These public displays are not constrained to techniques of obviously performative violence, such
as lynchings, that organizers “stage...for people to take in, notice, or experience firsthand” (Fujii,
2017, p. 661). All public violence fulfills the metaphor of performance by definition, whether or
not that performance occurs in the context of spaces that resemble a literal stage (Tilly, 2008).
Pogrom organizers use public acts like property destruction because they are visible assaults with
a broad audience, both in the moment of violence and in its aftermath (Bergmann, Hoffmann and
Smith, 2002). This group-selective violence aims to displace as well as to harm: it is a means of
reclaiming public space for the pogrom organizers, at the expense of the members of the targeted
community (Feldman, 1991). Organizers’ reliance on public violence as a trigger for collective
displacement is not unique to pogrom episodes: the psychological effects of terrorism (Hoffman,
1998) and paramilitary retribution during civil wars rely on a similar logic (Steele, 2017).
The symbolic logic of pogrom violence suggests that violence will take place where that symbol’s
social meaning is most salient—that is, where the groups that pogrom organizers seek to target
are most visible. Pogrom organizers select targets that will most effectively advance their message
of social exclusion. Mobilizing violence to displace or exclude a population with no visibility
or presence in a locality serves little strategic purpose. By contrast, visible populations offer a
compelling symbol of the social “other” that pogrom organizers seek to displace through violence.
Visible targets also aid pogrom organizers in mobilizing the “riot network” that enacts violence
(Brass, 1997). Like other social movements, pogrom organizers rely on social frames to motivate
and mobilize collective action. For movement organizers, participants, and observers alike, frames
define the basis for contention, its intended consequences, and the relationship between the actions
of the pogrom participants and the social outcomes that the organizers hope to accomplish. Pogrom
organizers activate local networks by arguing that successful violent action against a visible target
will strengthen the movement and cement its authority.
The most basic indicator of a population’s visibility is its relative size. A high concentration
of community members encourages the development of communal institutions that advance the
political and social interests of that community. These institutions include associational bodies
such as schools and sites of religious worship and ritual practice, and political groups like parties
and community advocacy groups. Where communities are less concentrated, they provide fewer
resources and fewer constituents to these organizational markers of public prominence.
Hypothesis 1. Areas with a relatively large population of the targeted group will experience higher
levels of group-selective violence after the onset of a pogrom.
3.2 Political Competition
Pogroms are also forms of political contestation. In many contexts, the outbreak of violence follows
pre-existing patterns of political and social competition. In India, Hindu and Muslim political
parties have used collective violence to appeal to different majority and minority constituencies.
The electoral incentives of competing parties shape the decision to mobilize or prevent costly
violence. In Wilkinson’s study of Hindu-Muslim riots, he finds a greater risk of ethnic rioting where
elections are more competitive, whereas pogrom organizers in Hindu or Muslim strongholds have
fewer incentives for collective violence (Wilkinson, 2004). Berenschot’s study of the 2002 pogrom in
Gujarat state also points to the interaction between these competitive incentives, existing systems
of patronage, and patterns of group-selective violence (Berenschot, 2011).
A second strain of competition-based arguments draws from “group-threat” or “power-threat”
theories of intergroup conflict (Spilerman, 1971; Olzak and Shanahan, 1996). These theories in-
terpret violence as a product of the social anxieties of the group responsible for organizing the
pogrom. “Group-threat” theories predict that violence will occur where and when pogrom or-
ganizers and participants perceive the targeted group as threatening to their social or economic
status. For proponents of group-threat theories, conditions that encourage the subjective percep-
tion of threat—rather than the objective possibility of collective deprivation—explain the onset
of violence (Quillian, 1995). Rapid changes in migration patterns or economic development often
precede these perceived threats to status. In Tulsa, Oklahoma, for example, the 1921 race riots
took place during a period of emergent Black economic success that left White Tulsa residents
concerned about the possibility of social displacement. In London’s Notting Hill neighborhood, the
race riots of 1958 occurred after local landlords evicted White tenants to make way for a growing
population of Afro-Caribbean residents (Pilkington, 1988). Kopstein and Wittenberg (2018) show
that anti-Semitic pogroms in contested areas of modern-day Poland and Ukraine in 1941 were most
likely to break out in localities with a higher concentration of pre-World War II support for Jewish
nationalist parties, whom Polish and Ukrainian nationalists saw as a threat to their own territorial
The best measure of violence as a function of political or social threat is the combined con-
centration of the pogrom organizers’ political opponents and the group targeted by the pogrom. If
the power-threat hypothesis holds, pogrom organizers will use their opponents’ political strength
as a pretense for violence against the targeted group.
Hypothesis 2. Areas with both higher levels of electoral support for the political opponents of
pogrom organizers and a relatively large population of the targeted group will experience higher
levels of group-selective violence after the onset of a pogrom.
3.3 Deep-Seated Hatreds
Pogroms also occur in a context of popular and elite animosity prejudices against the targeted
group. An essentialized version of this argument follows from Goldhagen’s (1996) controversial
study of the roots of the Holocaust in the cultural character of German anti-Semitism. Cultural
essentialists see violent mobilization as a persistent feature of some cultures, while others have no
similar propensity towards ethnonationalist violence (Connor, 1994). Essentialist arguments that
reify culture and theorize its relationship to violence as fixed in history have important historical
and logical flaws (Varshney, 2007). First, the ubiquity of ethnic hatred in societies in which mass
atrocities occur leads to the prediction that violence will be persistent and ever-present. If this
hypothesis holds, we should observe no instances of restraint or desistence in contexts of extreme
violence. The empirical record of mass-violence episodes is not consistent with this prediction.
Even in extreme cases like the Holocaust or the Rwandan genocide, individuals and groups that
share the ascriptive characteristics of the perpetrators perform heroic acts of resistance or rescue
(Tec, 1987; Braun, 2019; Fujii, 2009). Second, essentialist theories of deep-seated hatreds do not
account for the cultural pluralism of societies in which mass, group-selective violence occurs. In
Weimar Germany, for example, the momentary flourishing of pluralist politics suggested a place for
Jewish life and culture in competing conceptions of German society, long before the ascendancy of
the Nazi Party and its political allies was complete (Fritzsche, 1996).
A less essentialist version of the “deep-seated hatreds” argument follows from the idea that
pogroms are a costly form of collective action that involves the mobilization of like-minded par-
ticipants (Petersen, 2002). Pogroms involve the mobilization of people for whom acts of intimate
harm are not a regular form of social or political participation. Even if state license to violence
minimizes the social costs of collective violence, the emotional and psychological consequences of
enacting harm on others remain. Studies of genocide indicate that the decision to participate
in or refrain from violence against their neighbors often rests on the social relationships between
would-be perpetrators and the organizers of violence (McDoom, 2013; Luft, 2015; Fujii, 2009). A
higher concentration of people for whom animosity towards the targeted group is a salient political
preference lowers the social barriers to collective action. Violence after a pogrom’s onset should
occur where the ideological companions of pogrom organizers entertain a relatively high degree of
political and social support.
Hypothesis 3. Areas with higher levels of electoral support for ethnonationalist parties will expe-
rience higher levels of group-selective violence after the onset of a pogrom.
A last body of culturalist arguments emerges from the political science literature on conflict
recurrence. Studies of conflict recurrence indicate that places that have experienced violence in the
recent past are more likely to experience a new episode of mass violence than those without a recent
history of conflict (Das, 1990; Quinn, Mason and Gurses, 2007; Call, 2012). Ulfelder and Valentino
(2008) demonstrate that the recent occurrence of intrastate conflict indicates a future risk of mass
killing. In their study on the long-term effects of anti-Semitic pogroms in medieval Germany,
Voigtl¨ander and Voth (2012) show that German localities that experienced collective violence in
the 14th century were more likely to experience anti-Semitic violence during the Weimar and Nazi
periods. They attribute this empirical finding to the transmission of anti-Semitic attitudes across
Hypothesis 4. Areas with higher levels of previous violence against the targeted group will expe-
rience higher levels of group-selective violence after the onset of a pogrom.
3.4 Weak Civil Society
Although pogroms often represent a sharp departure from previous forms of abuse against the tar-
geted group, they often emerge from pre-existing associational configurations. Pogrom organizers
rely on social networks to distribute information about violent activities and encourage participa-
tion in violence (Horowitz, 2001; Scacco, 2008). Varshney’s (2002) research on Hindu-Muslim riots
in India points to the dual role of civic associations in either encouraging or restraining violence.
In areas in which Hindus and Muslims are relatively segregated, civic associations were less able
to stem the onset of ethnic violence. In communities in which these groups were more integrated,
by contrast, localities experienced less violence. Varshney attributes this finding to the possibility
of conflict resolution in integrated communities. Where Hindus and Muslims live together, they
establish “peace committees” that address contentious social and political topics without resorting
to violence. The absence of this conflict resolution architecture leaves more segregated communi-
ties more prone to outbreaks of violence. Comparative cases show that these cross-cutting peace
committees are a common means of violence prevention in conflict-prone societies (Kaplan, 2017;
Jose and Medie, 2015).
“Bridging” networks, which organize communities across various forms of social difference,
innoculate societies against the outcomes that their opposite, monoculture-reinforcing bonding
groups, encourage (Gittell and Vidal, 1998; Putnam, 2000). In contrast to “contact theories”
that predict the individual, attitudinal effects of interactions between members of different social
groups (Scacco and Warren, 2018; Paluck, Green and Green, 2018), “bridging” theories of violence
prevention emphasize the causal importance of collective association. At a national level, Bulutgil
(2016) finds that cross-cutting cleavages prevent ethnic cleansing by changing the political incentives
of ruling coalitions. These incentives also operate at the local level: where bridging organizations
are dominant, local leaders are more likely to oppose mobilization by pogrom organizers. These
organizations help to redefine the preferences of their members, who in participating in cross-
cutting associations accept that their collective interests align with those of social others. An
active member of an organization devoted to encouraging cross-cultural collaboration, for example,
has social incentives to protect and advocate for members of other cultural groups that a member
of a “bonding” organization devoted to the preservation of a specific cultural identity does not have
(Braun, 2019).
Popular support for political pluralism is a useful indicator of the strength of bridging or-
ganizations capable of opposing political violence. Pluralist networks should protect against the
outbreak of violence; risks of violence are greater where those networks are weaker.
Hypothesis 5. Areas with higher levels of electoral support for pluralist parties will experience
lower levels of group-selective violence after the onset of a pogrom.
4 Research Design
I examine these explanations for violence in the case of the Kristallnacht pogrom (“Night of Broken
Glass”, sometimes also called Reichskristallnacht or Novemberpogrom), which senior Nazi Party
officials incited against the Jewish population of Germany and Austria in November 1938. On
November 9th and 10th, Nazi security forces, paramilitary groups, and civilian party members
organized a rapid series of attacks on Jewish civilians, institutions, and businesses that spanned
the full territory of the Third Reich. For German Jews, the Nazi regime’s quick and brutal assault
on their increasingly marginalized communities was a clear harbinger of the subsequent six years
of genocidal violence during the Nazi Holocaust.
I use a multivariate regression analysis of variations in violence at the municipal level to assess
competing explanations for the Kristallnacht pogrom’s spread. Statistical regression techniques
assess whether a large number of observations with common characteristics tend towards a common
outcome. The multivariate design of the following regression models addresses the possibility that
other social or political characteristics of each municipality might account for violence outcomes.
In the following section, I recount the macro- and meso-level processes that led to the national
wave of pogrom violence. Following this qualitative description, I describe the outcome variables,
explanatory variables, and covariates that make up my regression models.
4.1 The Kristallnacht Pogrom, November 1938
The hour-by-hour decision process that preceded the Kristallnacht pogrom is well-documented.
The proximate trigger for the Kristallnacht pogrom was the assassination of a German diplomat,
Ernst vom Rath, by Herschel Grynzspan, a Polish Jew, in the morning of November 7th. Grynzs-
pan’s acted in protest against the Nazi regime’s forced deportation of Polish-Jewish residents from
Germany just weeks before. Within hours of the events in Paris, Nazi propagandists transformed
vom Rath’s assassination from the actions of a singular rebel to apparent evidence of an interna-
tional Jewish conspiracy against the German nation. That evening, officials distributed reports of
the conspiracy to local German press shops and, by extension, local party functionaries (Heiber,
1957). These anti-Semitic accusations had no factual basis—instead, the obvious purpose of the
information was to“operate not rationally but viscerally” (Steinweis, 2009, p. 21), so as to stoke
popular enthusiasm for the conspiratorial frames of Nazi leadership.
Local Nazi officials organized violence in select towns as early as the night of November 7th
(Kropat, 1988), but the nationwide pogrom did not begin in earnest until November 9th. Earlier
that day, Nazi leadership had descended on the Bavarian city of Munich, the symbolic capital of
the Nazi movement, to commemorate the 15th anniversary of the Party’s failed putsch against the
Bavarian state government. The year 1938 had seen a growing number of instances of large-scale
physical violence against Jewish individuals and businesses, and constrained instances of direct anti-
Semitic violence during the previous five years received little censure from Nazi leaders. Hesitation
about Germany’s international reputation, however, made Nazi leadership hesitant to launch a
more coordinated assault on the country’s Jewish population (Longerich, 2010). The strategic
calculus of the Nazi regime changed, however, after the Reich’s successful annexation of Austria in
spring 1938. Hawkish members of Adolf Hitler’s inner circle agitated throughout the fall for a more
ambitious policy of anti-Semitic abuse; on November 9th, Hitler assented to a broad onslaught of
violence against German Jews.
After receiving Hitler’s approval, Goebbels instructed a group of Nazi leaders in Munich to en-
courage violence against Jewish businesses and synagogues throughout the country. Reconstructing
the early hours of the pogrom from a 1939 Party investigation and the recollections of multiple Nazi
officials including Goebbels himself, Steinweis describes the onset as “hectic and more than a little
chaotic”: “The improvised and messy process for disseminating the pogrom authorization...helps to
underscore the fact that Hitler and Goebbels had not decided upon the nationwide pogrom earlier
than the night of November 9” (Steinweis, 2009, p. 47–8). The improvisation, however, meant
that some local constituencies engaged in violence in excess of the severity that Nazi leaders had
intended, while others fell short of its brutal intensity. The main direct perpetrators of the violence
were members of the Party’s Sturmabteilung (SA) paramilitary units, as well as party members and
local youth groups who joined in the violence towards the end of November 10th (Siemens, 2017).
Goebbels’s vague orders meant that these perpetrators deployed a diverse set of violent tech-
niques against their Jewish targets. In addition to destroying Jewish synagogues, prayer houses,
and businesses, paramilitary forces and party members killed, harassed, and beat their Jewish
neighbors. The violence included the destruction of Jewish materials of high financial and spiritual
value. A wave of arrests by the Nazi secret police accompanied the episode of collective violence,
resulting in the first large-scale deportations to existing concentration camps throughout Germany.
There were exceptions: although violence was the order of the day, some non-Jewish Germans re-
sisted violence, aided their Jewish counterparts, and refused to participate in anti-Semitic attacks
(Steinweis, 2009). As discussed below, approximately one-third of all German synagogues bore no
evidence of damage.
The Kristallnacht pogrom provides a rich case for empirical inquiry about the causes of spatial
variation in pogrom violence. In theoretical terms, the proximate cause of the violence—Hitler’s
decision to incite a national wave of anti-Semitic attacks in response to vom Rath’s assassination—
was exogenous to the spread of the pogrom at the municipal level. There was no regimented chain of
command that coordinated the violence, and the chaotic process by which party officials in Munich
communicated Goebbels’s orders to local party officials on November 9th suggests that no one
locality was more likely to receive pogrom orders than another. Party officials had a greater deal
of local agency in the mobilization of violence than conventional interpretations of Nazi violence
suggest. These exogenous orders and the relative brievity of the episode give greater confidence that
the common features of locations in which violence took place, and not unobserved characteristics
of the episode’s organization, account for its spread.
The events of Kristallnacht share multiple features with pogrom episodes in other global con-
texts. The episode took place under conditions of state license, rather than collapse. German offi-
cials might have attempted to constrain the violence against Jewish communities; the Nazi regime’s
unlikely prosecution of select Kristallnacht perpetrators for acts of rape and plunder against their
Jewish neighbors indicates some measure of discretion about the scale and techniques of anti-
Semitic assault (McKale, 1973). The possibility of constraint contrasts to an episode like the wave
of pogroms in Eastern Europe just two years later, which occurred amid multiple competing claims
for territorial sovereignty over Poland’s eastern borderlands. Additionally, officials with extra-local
authority—in this case, senior Nazi Party officials—exercised only limited control over the coordi-
nation and targets of violence at a local level. The degree of agency that local communities and
party officials exhibited during the Kristallnacht pogrom aligns with corollary concepts like “riot
networks” in India (Brass, 1997) or “lynching regimes” in the United States (Seguin and Rigby,
2019), in which local groups have some autonomous space for improvisation, but act in conjunction
with the pogrom’s central organizing logic. Here, it is important to distinguish our knowledge of
the mechanized regime of anti-Semitic violence that facilitated the events of the Holocaust, from
the contested autocracy that ruled Nazi Germany during the prewar period. Pre-1939 Germany
was a single-party state, but one in which local party officials had a significant say in the occurrence
and severity of political violence.
4.2 Outcome Variables
Destruction of synagogues: I use multiple municipal-level measures of attacks on Jewish synagogues
and prayer houses to assess group-selective violence during the Kristallnacht pogrom. The common
social purpose of synagogues and senior Nazi instructions to damage them make attacks on these
buildings a clear indicator of group-selective violence during the pogrom. For both German Jews
and non-Jews around them, the synagogue was the most obvious institution of Jewish civil and
religious society (Brenner, 1998). In Weimar Germany, the synagogue was a house of worship
for ritually observant Jews and a central gathering space for Jewish associations across multiple
denominations. The social and symbolic functions of the synagogue persisted as official Nazi anti-
Semitism grew more entrenched. Theoretically, attacks on synagogues during the Kristallnacht
pogrom were equally likely in all German localities: physical attacks on Jewish synagogues were a
clear directive from senior Nazi leadership, whereas lethal and non-lethal assaults against Jewish
individuals were not.
I rely on a new geocoded dataset of Jewish synagogues and prayer houses based on the memo-
rialization work of the Jerusalem-based Beth Ashkenaz Synagogue Memorial Project (Synagogue
Memorial Beth Ashkenaz, 2003). To commemorate the 65th anniversary of the pogrom in 2003,
the memorial project published a list of synagogues and prayer houses that existed in Germany
prior to the events of Kristallnacht. For sites attacked or destroyed during the Kristallnacht
pogrom, the list briefly describes the conditions surrounding the attack. An updated project
website (, released in 2019, contains additional information
about the history, associational life, and ritual practices of the Jewish communities in towns that
experienced violence during the Kristallnacht pogrom.
The list includes the city, state, and current-to-2003 address of each religious site. I use the
Google Maps (ggmap) package in the statistical software R to identify the spatial point (latitude
and longitude) of each synagogue site. Due to scarce information, the list excludes or provides
incomplete information on sites in the provinces of eastern Silesia, East Prussia, and Pomerania.
To enable systematic comparison with Weimar-era political and social data, I also exclude syn-
agogues and prayer houses in these provinces from the dataset. The geocoded dataset includes
2,238 observations, each representing an individual synagogue or prayer house active in Germany
at some date prior to the Kristallnacht pogrom. The Beth Ashkenaz list describes 1,890 institutions
as still active in 1938. The list describes 1,253 (66 percent) of the active synagogues as having been
damaged or destroyed by local pogrom participants, while providing no evidence of attacks against
637 (34 percent) of them. Figure 1 displays the spatial coordinates of attacked and non-attacked
synagogues based on the list, plotted on a map of the 1933 boundaries of the Weimar Republic.
Figure 1: Synagogues Attacked and Not Attacked during the Kristallnacht Pogrom
10 15 20
Not attacked
4.3 Explanatory Variables
The hypotheses described in Section 3 point to five explanatory variables that may account for
spatial variation in synagogue attacks. Each of the explanatory variables describes a political or
social feature of the Weimar Republic, which ruled Germany from 1918 to the Nazi Party’s rise
to power in 1933. I infer the social characteristics and collective political preferences of German
localities in 1938 from these pre-1933 data. Although the rise of the Nazi Party was an obvious
shock to the distribution of political power in German society, the demographic characteristics of
communities across Germany remained relatively constant until the events of Kristallnacht—the
first instance of acute nationwide violence under Nazi rule—and the beginning of World War II in
late 1939.
These measures are available because of the extensive resources that German social scientists,
especially those affiliated with the Leibniz-Institut f¨ur Sozialwissenschaften (GESIS) in K¨oln, Ger-
many, have devoted to collating and disseminating Weimar-era social and political data for scholarly
use (H¨anisch, 1988; Falter, 1991).
Jewish population share in 1933 : For the relative size of the targeted group, I use the Jewish
population share as calculated in the 1933 census. The population size of Germany’s Jewish mi-
nority was associated with other communal characteristics for which we lack systematic data, such
as the scale of Jewish associations, the proliferation of sites of religious worship, and Jewish busi-
ness activities (Volkov, 2006). Where Jews were more populous, they were better able to establish
visible institutions that put their civic strength on display.
The source of this measure merits some discussion, not least because the 1933 census was
among the new Nazi regime’s first attempts to weave the ideology of racial eugenics into German
state institutions (Luebke and Milton, 1994). As with any data generated by the Nazi Party, the
use of this data presents obvious ethical and empirical quandaries (Post, 1991). Might German
census-takers have overestimated the size of the country’s Jewish population in an attempt to
justify a wider anti-Semitic crackdown? In fact, historical evidence suggests that the 1933 census
may have undercounted Germany’s Jews. The Nazi regime conducted the census before the 1935
Nuremberg laws that established a blood-lineage standard for anti-Semitic discrimination. As a
result, the census-takers mainly recorded the country’s visibly observant population rather than
its more assimilated membership (Luebke and Milton, 1994). Given that the Kristallnacht pogrom
took place after the regime established the Nuremberg standard for Jewish lineage, the 1933 census
provides a conservative estimate of the size of targeted Jewish communities in each municipality.
Vote share of the Communist Party in the September 1930 federal election: For electoral
support for the political opponents of the pogrom organizers, I use the Communist Party (KPD)
vote share in the September 1930 election. The September 1930 election is the last open federal
election in Weimar Germany for which municipal-level data are available. The KPD was the Nazi
Party’s diametric opponent, both at the polls and in the pitched streetfights that accompanied
democratic politics throughout the Weimar era (Bessel, 1984). After the Nazi Party’s rise to power,
Communist activists, intellectuals, and party members were the first targets of Nazi violence. By
1934, Nazi paramilitary groups and German security forces had succeeded in repressing almost all
overt Communist political activity throughout Germany.
Despite the extent of anti-Communist violence in the early Nazi years, Nazi rhetoric through-
out the prewar period justified harsh security measures as responses to both local and global Com-
munism. This propaganda often relied on the anti-Semitic myth of “Judeo-Communism,” which
associated an international Jewish conspiracy with the global diffusion of Communist ideology de-
spite scant evidence of significant Jewish control over KPD activities (Friedlander, 1997). To assess
political competition against perceived Jewish Communists as a potential explanation for violence
during the Kristallnacht pogrom, I estimate the multiplicative effect of combined electoral support
for the KPD and relatively high Jewish population share on synagogue attacks.
Vote share of the anti-Semitic coalition: For electoral support for ethnonationalist groups, I
use the combined vote share of the two main ethnonationalist parties, the Nazis and the German
National People’s Party (DNVP), that contested the September 1930 election. In the late Weimar
period, both the Nazi Party and the DNVP jockeyed for support among right-wing nationalist
constituencies (Chanady, 1967). Both employed explicitly anti-Semitic platforms and promoted
themselves as opponents of the pluralist Weimar order (Wildt, 2014).
Pogroms in the 14th century and 1920s: For past violence against the targeted group, I use
historical patterns of pogrom violence against German Jews. I estimate the effects of both medieval
and more contemporaneous instances of anti-Semitic violence, following Voigtl¨ander and Voth’s
(2012) finding that the occurrence of pogroms during the 14th-century Black Death plague is
associated with higher levels of Weimar- and Nazi-era anti-Semitism across multiple measures.
Vote share of the Social Democratic Party : For electoral support for pluralist parties, I use the
Social Democratic Party (SPD) vote share during the September 1930 election. Throughout the
Weimar period, the SPD adopted the combined mantles of republicanism and political pluralism
(Wildt, 2014). Aside from the KPD, the SPD was the only consistent opponent of the widespread
anti-Semitic rhetoric among right-wing German parties. They also condemned prominent episodes
of anti-Semitic violence during the Weimar era.
4.4 Covariates
I include two covariates, population size and the predicted strength of local radio signals, that are
typically associated with pogrom violence but to which I do not assign causal significance.
Population: The relative ease of mass mobilization in urban centers and the relative hetero-
geneity of most urban communities means that pogroms often occur in urban areas (Tambiah,
1997; Varshney, 2002). To account for the urban character of pogrom violence, I control for the
natural log of each municipality’s population according to the 1933 census. Using the log of popula-
tion, rather than its non-exponentiated value, addresses the possibility of a non-linear relationship
between population and violence.
Radio access: Organizers of mass-violent episodes rely on mass-communication technologies
to facilitate violence, stoke conspiratorial rumors, and encourage popular resentment against the
targeted group (Horowitz, 2001; Myers, 2000; Straus, 2007-12-01; Yanagizawa-Drott, 2014). Radio
was a critical means of Nazi mass communication with the German populace and, following the
Nazi Party’s rise to power, an effective platform for the widespread dissemination of anti-Semitic
propaganda (Welch, 2002). At the onset of the Kristallnacht pogrom, Nazi officials used a com-
bination of telephones, press wires, and the national radio service to spread reports about Ernst
vom Rath’s assassination (Steinweis, 2009). To account for the potential influence of radio on the
coordination of violence during the pogrom, I multiply Adena et al.’s (2015) calculation of predicted
radio signal strength in 1938 by electoral support for the anti-Semitic coalition of the Nazi Party
and the DNVP.
I also use province fixed effects, which account for the unobserved characteristics of some
regions that may have influenced the occurrence, spread, or intensity of Nazi perpetrators’ violence.
I display the summary statistics for the outcome variables, explanatory variables, and covari-
ates in Table 1.
Table 1: Summary Statistics for Kristallnacht Variables
Statistic N Mean St. Dev. Min Max Source
Attacks (binary) 1,131 0.732 0.443 0 1 New dataset
Attacks (count) 1,131 1.108 1.776 0 42 New dataset
Attacks (percent) 1,045 0.709 0.406 0.000 1.000 New dataset
Jewish share, 1933 922 0.024 0.034 0.0001 0.377 anisch, 1988
KPD, 1930 1,131 0.076 0.073 0.002 0.427 anisch, 1988
Anti-Sem., 1930 1,131 0.223 0.117 0.020 0.850 H¨anisch, 1988
Pogroms, 1349 1,027 0.203 0.402 0.000 1.000 Voigtl¨ander and Voth, 2012
Pogroms, 1920s 1,108 0.026 0.160 0.000 1.000 Voigtl¨ander and Voth, 2012
SPD, 1930 1,131 0.202 0.128 0.006 0.619 H¨anisch, 1988
Radio strength, 1938 1,131 24.799 9.648 5.411 83.045 Adena et al., 2015
Pop. 1933 (log) 1,108 8.250 1.586 4.927 15.308 anisch, 1988
5 Findings
I estimate three linear regression models, one for each measure of the outcome variable. I display the
standardized results of the models in Table 2. I also visualize the standardized coefficients and their
95-percent confidence intervals in Figure 2. I estimate a logit regression and a Poisson regression
model, which correspond to the binary and right-skewed characteristics of the “occurrence” and
“count” measures of the outcome variable, respectively. I present the odds ratios associated with
the logit and Poisson regression models in Table 3. The results of the linear models are robust to
the non-linear model specifications.
The regression results provide strong evidence that pogroms are most severe where the popu-
lation targeted by the organizers is most visible. During the Kristallnacht pogrom, the effect of the
relative size of the Jewish population on the occurrence and relative severity of pogrom violence
(Model 1 and Model 3, respectively) was both substantively and statistically distinguishable from
zero. Pogrom organizers targeted a higher number of synagogues (Model 2) in municipalities that
were also pre-1933 KPD strongholds. The independent effect of the relative share of Jews on the
number of attacked synagogues is substantively small and statistically insignificant at α= 0.05
(p= 0.069). The effect of the interaction term suggests that the myth of Judeo-Communism was
a durable pretense for Nazi violence, despite the regime’s effective and systematic campaign of
anti-Communist repression during the five years prior to the Kristallnacht pogrom. However, the
multiplicative effect of Weimar-era support for Communism—an indicator of the political competi-
tion hypothesis—does not explain either the occurrence or relative severity of anti-Semitic violence
during the Kristallnacht pogrom.
There is no evidence that a high concentration of anti-Semitic sentiment increased anti-Semitic
violence during the Kristallnacht pogrom. Surprisingly, attacks on synagogues were less common
and less severe where the anti-Semitic coalition of the Nazi Party and the DNVP entertained
less political support. Neither medieval-era pogroms nor pogroms during the Weimar period had
a statistically or substantively significant effect on any of the three municipal-level measures of
synagogue attacks. The effect of electoral support for the SPD was similarly indistinguishable
from zero, indicating that political pluralism did little to constrain anti-Semitic violence after the
episode’s onset.
Nazi access to radio did not have a substantively or statistically significant effect on violence
during the Kristallnacht pogrom. Nazi officials’ predominant use of private telephone communi-
cations in the coordination of violence, rather than public media reporting associated with radio
communications, may explain this null effect. The statistically and substantively significant ef-
fect of logged population size in all three models confirms that anti-Semitic violence during the
Kristallnacht pogrom was a predominantly urban phenomenon.
Table 2: Regression Results for Kristallnacht Models (standardized)
Synagogue attacks
Binary Count Percent
(1) (2) (3)
Jewish pop. share (1933) 0.109∗∗∗ 0.0740.236∗∗∗
(0.023) (0.041) (0.055)
t = 4.693 t = 1.814 t = 4.281
Jews (1933) KPD share (1930) 0.020 0.133∗∗∗ 0.094
(0.023) (0.040) (0.053)
t = 0.900 t = 3.357 t = 1.781
KPD share (1930) 0.012 0.059 0.071
(0.021) (0.037) (0.050)
t = 0.570 t = 1.594 t = 1.411
Anti-Sem. share (1933) 0.070∗∗ 0.017 0.149
(0.034) (0.060) (0.081)
t = 2.048 t = 0.285 t = 1.842
Pogroms, 1349 0.023 0.007 0.041
(0.016) (0.028) (0.038)
t = 1.405 t = 0.252 t = 1.081
Pogroms, 1920s 0.002 0.035 0.014
(0.014) (0.025) (0.035)
t = 0.143 t = 1.387 t = 0.407
SPD share (1930) 0.004 0.041 0.004
(0.018) (0.031) (0.042)
t = 0.252 t = 1.321 t = 0.084
Anti-Sem. (1930) Radio (1938) 0.044 0.074 0.098
(0.041) (0.071) (0.097)
t = 1.074 t = 1.036 t = 1.006
Radio strength (1938) 0.045 0.157∗∗∗ 0.099
(0.031) (0.055) (0.074)
t = 1.426 t = 2.855 t = 1.334
1933 pop. (log) 0.115∗∗∗ 0.357∗∗∗ 0.101
(0.022) (0.039) (0.053)
t = 5.135 t = 9.075 t = 1.903
Constant 0.756∗∗∗ 0.066 0.252∗∗
(0.046) (0.082) (0.113)
t = 16.294 t = 0.803 t = 2.218
Observations 827 827 774
R20.129 0.263 0.113
Note: p<0.1; ∗∗p<0.05; ∗∗∗ p<0.01
Figure 2: Regression Results for Kristallnacht Models (standardized)
1933 pop. (log)
Pogroms, 1920s
Pogroms, 1349
SPD share (1930)
Anti−Sem. (1930) * Radio (1938)
Radio strength (1938)
Anti−Sem. share (1933)
KPD share (1930)
Jews (1933) * KPD share (1930)
Jewish pop. share (1933)
Table 3: Regression Results for Kristallnacht Models (standardized, non-linear)
Synagogue attacks
Binary Count
logistic Poisson
(1) (2)
Jewish pop. share (1933) 3.012∗∗∗ 1.090
(0.681) (0.056)
Jews (1933) * KPD share (1930) 0.868 1.151∗∗∗
(0.152) (0.053)
KPD share (1930) 1.142 0.953
(0.198) (0.042)
Anti-Sem. share (1933) 0.6590.996
(0.146) (0.086)
Pogroms, 1349 1.139 1.006
(0.137) (0.035)
Pogroms 1920s 1.009 0.963
(0.131) (0.030)
SPD share (1930) 0.974 0.957
(0.109) (0.041)
Anti-Sem. (1930) * Radio (1938) 1.283 0.926
(0.346) (0.087)
Radio strength (1938) 0.786 1.168∗∗
(0.166) (0.081)
1933 pop. (log) 2.753∗∗∗ 1.503∗∗∗
(0.511) (0.067)
Constant 4.075∗∗∗ 0.959
(1.236) (0.105)
Observations 827 827
Log Likelihood 372.654 1,036.581
Note: p<0.1; ∗∗p<0.05; ∗∗∗ p<0.01
6 Conclusion
This paper sought to test prevailing explanations for why pogroms lead to violence in some localities,
but not others. I defined pogroms as a particular form of group-selective violence involving multiple,
public violent acts by an informal group. I introduced a new geocoded dataset, based on work by
a Jewish memorialization project, that documents variations in local perpetrators’ attacks against
Jewish synagogues and prayer houses in Germany during the Kristallnacht pogrom in November
1938. I used this dataset and municipal-level data about the political and social characteristics of
Weimar Germany to test four competing theories for spatial variation during pogroms, drawn from
the literatures on violent conflict, inter-group relations, and contentious politics.
I found that pogrom organizers use group-selective violence during pogroms for symbolic pur-
poses, to communicate and advance a political agenda that excludes targeted communities from
the existing social order. This finding underscores the fundamental tragedy of the German Jewish
experience of the Holocaust: German Jewish communities were most vulnerable to Nazi violence
where they had thrived most during the Weimar period. I also found that pogrom organizers use
group-selective violence as a form of political competition against their political opponents. I found
no evidence that group-selective violence during pogroms is associated with a high concentration of
ethnonationalist sentiment, nor that communal support for political pluralism is a bulwark against
These findings have important theoretical, empirical, and methodological implications for
scholars of political violence. Pogroms are the result of mobilization “from above”—by senior
or national political officials, party apparatchiks, and senior officials of formal and informal se-
curity forces—and “from below,” by local movement actors. When senior officials incite violence
against particular social groups, local pogrom organizers engage in that group-selective violence
where the targeted groups are most visible. These empirical patterns suggest that scholars and
practitioners seeking to explain and prevent pogrom violence should understand pogroms and their
organizers as social movements engaged in strategic behavior, rather than as the natural outgrowth
of fixed patterns of intergroup relations or ethnic prejudice. Additionally, the findings suggest that
the political violence literature overstates the divide between rationalist and culturalist origins
of violence (Brubaker and Laitin, 1998). The violence of the Kristallnacht pogrom advanced the
strategic interests of senior Nazi officials and local party organizers, allowing both to further cement
their hold over German society. But we cannot understand the spatial patterns of that violence
without reference to the social status and practices of the pogrom’s Jewish targets. The interaction
between these characteristics mean that further inquiry about pogroms requires a methodologically
syncretic approach that combines comparative empirical analysis with interpretive studies of social
practices, relations, and meanings.
My findings also provoke new questions about the perpetrators and victims of pogrom violence.
The findings point to variation in group-selective violence at the municipal level. Do the same
factors explain the individuals, households, or small groups that perpetrators are most likely to
target in particular localities? Once perpetrators decide to participate in a pogrom event, what
explains why they opt for one repertoire of violent behavior, such as killing or property destruction,
versus or in addition to others? For those groups that pogrom organizers target, what social
practices mark the visibility of their communities other than a relatively large concentration of
group members? Do different forms of visibility leave individuals and communities more or less
likely experience violence? As pogroms persist in diverse places such as Myanmar, Sri Lanka, and
Nigeria, understanding the relationship between these factors and pogrom dynamics will contribute
to growing knowledge about and efforts to prevent this hateful violence.
Adena, Maja, Ruben Enikolopov, Maria Petrova, Veronica Santarosa and Ekaterina Zhuravskaya.
2015. “Radio and the Rise Of the Nazis in Prewar Germany.” Quarterly Journal of Economics
Berenschot, Ward. 2011. “The Spatial Distribution of Riots: Patronage and the Instigation of
Communal Violence in Gujarat, India.” 39(2):221–230.
Bergmann, Werner, Christhard Hoffmann and Helmut W. Smith. 2002. Exclusionary Violence:
Antisemitic Riots in Modern German History. Ann Arbor, MI: University of Michigan Press.
Bessel, Richard. 1984. Political Violence and the Rise of Nazism: The Storm Troopers in Eastern
Germany, 1925-1934. Yale University Press.
Bourdieu, Pierre. 1980. The Logic of Practice. Stanford University Press.
Bourdieu, Pierre. 1993. Language and Symbolic Power. Cambridge, MA: Harvard University Press.
Brass, Paul R. 1997. Theft of an Idol: Text and Context in the Representation of Collective Violence.
Princeton University Press.
Brass, Paul R. 2003. The Production of Hindu-Muslim Violence in Contemporary India. University
of Washington Press.
Brass, Paul R., ed. 1996. Riots and Pogroms. Springer.
Braun, Robert. 2019. Protectors of Pluralism: Religious Minorities and the Rescue of Jews in the
Low Countries during the Holocaust. Cambridge, UK: Cambridge University Press.
Brenner, Michael. 1998. The Renaissance of Jewish Culture in Weimar Germany. New Haven, CT:
Yale University Press.
Brubaker, Rogers and David D. Laitin. 1998. “Ethnic and Nationalist Violence.” Annual Review
of Sociology 24(1):423–452.
Bulutgil, H. Zeynep. 2016. The Roots of Ethnic Cleansing in Europe. Cambridge, UK: Cambridge
University Press.
Call, Charles T. 2012. Why Peace Fails: The Causes and Prevention of Civil War Recurrence.
Washington, DC: Georgetown University Press.
Chanady, Attila. 1967. “The Disintegration of the German National Peoples’ Party 1924-1930.”
The Journal of Modern History 39(1):65–91.
Connor, Walker. 1994. Ethnonationalism: The Quest for Understanding. Princeton, NJ: Princeton
University Press.
Das, Veena, ed. 1990. Mirrors of Violence: Communities, Riots and Survivors in South Asia.
Oxford, UK: Oxford University Press.
Ellsworth, Scott. 1992. Death in a Promised Land: The Tulsa Race Riot of 1921. LSU Press.
Falter, Jurgen W. 1991. Hitlers Wahler. C.H. Beck.
Feldman, Allen. 1991. Formations of Violence: The Narrative of the Body and Political Terror in
Northern Ireland. Chicago: University of Chicago Press.
Friedlander, Saul. 1997. Nazi Germany and the Jews: Volume 1: The Years of Persecution 1933-
1939. New York: Harper Collins.
Fritzsche, Peter. 1996. “Did Weimar Fail?” The Journal of Modern History 68:629–656.
Fujii, Lee Ann. 2009. Killing Neighbors: Webs of Violence in Rwanda. Cornell University Press.
Fujii, Lee Ann. 2013. “The Puzzle of Extra-Lethal Violence.” Perspectives on Politics 11(2):410–
Fujii, Lee Ann. 2017. “‘Talk of the town’: Explaining pathways to participation in violent display.”
Journal of Peace Research 54(5):661–673.
Ghassem-Fachandi, Parvis. 2012. Pogrom in Gujarat: Hindu Nationalism and Anti-Muslim Vio-
lence in India. Princeton, NJ: Princeton University Press.
Gittell, Ross and Avis Vidal. 1998. Community Organizing: Building Social Capital as a Develop-
ment Strategy. Thousand Oaks, CA: SAGE Publications.
Goldhagen, Daniel Jonah. 1996. Hitler’s Willing Executioners: Ordinary Germans and the Holo-
caust. Knopf Doubleday Publishing Group.
Gross, Jan T. 2001. Neighbors: The Destruction of the Jewish Community in Jedwabne, Poland.
Princeton University Press.
Guti´errez-San´ın, Francisco and Elisabeth Jean Wood. 2017. “What Should We Mean by “Pattern of
Political Violence”? Repertoire, Targeting, Frequency, and Technique.” Perspectives on Politics
Heiber, Helmut. 1957. “Der Fall Gr¨unspan.” Vierteljahrshefte f¨ur Zeitgeschichte pp. 134–172.
anisch, Dirk. 1988. “Inhalt und Struktur der Datenbank Wahl- und Sozialdaten der Kreise und
Gemeinden des Deutschen Reiches von 1920 bis 1933.” Historical Social Research / Historische
Sozialforschung 14(1):39–67.
Hobsbawm, Eric and George Rud´e. 1969. Captain Swing. UK: Lawrence & Wishart.
Hoffman, Bruce. 1998. Inside Terrorism. New York: Columbia University Press.
Hoover Green, Amelia. 2018. The Commander’s Dilemma: Violence and Restraint in Wartime.
Cornell University Press.
Horowitz, Donald L. 2001. The Deadly Ethnic Riot. Berkeley, CA: University of California Press.
Jose, Betcy and Peace A. Medie. 2015. “Understanding Why and How Civilians Resort to Self-
Protection in Armed Conflict.” International Studies Review 17.
Kaplan, Oliver. 2017. Resisting War: How Communities Protect Themselves. Cambridge, UK:
Cambridge University Press.
King, Charles. 2012. “Can There Be a Political Science of the Holocaust?” Perspectives on Politics
Klier, John Doyle and Shlomo Lambroza, eds. 1992. Pogroms: Anti-Jewish Violence in Modern
Russian History. Cambridge, UK: Cambridge University Press.
Kopstein, Jeffrey S. and Jason Wittenberg. 2018. Intimate Violence: Anti-Jewish Pogroms on the
Eve of the Holocaust. Ithaca, NY: Cornell University Press.
Krause, Jana. 2018. Resilient Communities: Non-Violence and Civilian Agency in Communal War.
Cambridge University Press.
Kropat, Wolf-Arno. 1988. Kristallnacht in Hessen: der Judenpogrom vom November 1938. Schriften
der Kommission f´
’ur die Geschichte der Juden in Hessen.
Longerich, Peter. 2010. Holocaust: The Nazi Persecution and Murder of the Jews. Oxford University
Lower, Wendy. 2011. “Pogroms, mob violence and genocide in western Ukraine, summer 1941:
varied histories, explanations and comparisons.” 13(3):217–246.
Luebke, David Martin and Sybil Milton. 1994. “Locating the victim: An overview of census-
taking, tabulation technology and persecution in Nazi Germany.” IEEE Annals of the History of
Computing 16(3).
Luft, Aliza. 2015. “Toward a Dynamic Theory of Action at the Micro Level of Genocide: Killing,
Desistance, and Saving in 1994 Rwanda.” Sociological Theory 33(2):148–172.
McDoom, Omar. 2013. “Anti-social capital: A profile of Rwandan genocide perpetrators’ social
networks.” Journal of Conflict Resolution 58:865–893.
McKale, Donald. 1973. “A Case of Nazi ”Justice”: The Punishment of Party Members Involved in
the Kristallnacht, 1938.” 35(3):228–238.
Myers, Daniel J. 2000. “The Diffusion of Collective Violence: Infectiousness, Susceptibility, and
Mass Media Networks.” American Journal of Sociology 106(1):173–208.
Olzak, Susan and Suzanne Shanahan. 1996. “Deprivation and Race Riots: An Extension of Spiler-
man’s Analysis.” 74(3):931–961.
Paluck, Elizabeth Levy, Seth A. Green and Donald P. Green. 2018. “The contact hypothesis
re-evaluated.” Behavioural Public Policy pp. 1–30.
Petersen, Roger D. 2002. Understanding Ethnic Violence: Fear, Hatred, and Resentment in
Twentieth-Century Eastern Europe. Cambridge University Press.
Pilkington, Edward. 1988. Beyond the mother country: West Indians and the Notting Hill white
riots. Tauris.
Piven, Frances Fox and Richard A. Cloward. 1977. Poor People’s Movements: Why They Succeed,
how They Fail. New York: Pantheon Books.
Post, Stephen G. 1991. “The Echo of Nuremberg: Nazi Data and Ethics.” Journal of Medical
Ethics 17:42–44.
Putnam, Robert D. 2000. Bowling Alone: Collapse and Revival of American Community. New
York: Simon & Schuster.
Quillian, Lincoln. 1995. “Prejudice as a Response to Perceived Group Threat: Population Com-
position and Anti-Immigrant and Racial Prejudice in Europe.” American Sociological Review
Quinn, J. Michael, T. David Mason and Mehmet Gurses. 2007. “Sustaining the Peace: Determi-
nants of Civil War Recurrence.” International Interactions 33(2):167–193.
Scacco, Alexandra. 2008. Who Riots? Explaining Individual Participation in Ethnic Violence PhD
thesis Columbia University.
Scacco, Alexandra and Shana S. Warren. 2018. “Can Social Contact Reduce Prejudice and Dis-
crimination? Evidence from a Field Experiment in Nigeria.” American Political Science Review
Seguin, Charles and David Rigby. 2019. “National Crimes: A New National Data Set of Lynchings
in the United States, 1883 to 1941.” Socius 5.
Sidel, John T. 2006. Riots, Pogroms, Jihad: Religious Violence in Indonesia. Ithaca, NY: Cornell
University Press.
Siemens, Daniel. 2017. Stormtroopers: A New History of Hitler’s Brownshirts. New Haven, CT:
Yale University Press.
Snow, David A. and Robert D. Benford. 1992. Master Frames and Cycles of Protest. In Frontiers
in Social Movement Theory, ed. Aldon D. Morris and Carol McClurg Mueller. Yale University
Spilerman, Seymour. 1971. “The Causes of Racial Disturbances: Tests of an Explanation.” Amer-
ican Sociological Review (3):427–442.
Steele, Abbey. 2017. Democracy and Displacement in Colombia’s Civil War. Cornell University
Steinweis, Alan E. 2009. Kristallnacht 1938. Harvard University Press.
Straus, Scott. 2007-12-01. “What Is the Relationship between Hate Radio and Violence? Rethinking
Rwanda’s “Radio Machete”.” Politics & Society 35(4):609–637.
Synagogue Memorial Beth Ashkenaz. 2003. List of Germany’s synagogues of the 20th Century.
Synagogue Memorial Beth Ashkenaz.
Tajima, Yuhki. 2014. The Institutional Origins of Communal Violence: Indonesia’s Transition
from Authoritarian Rule. Cambridge University Press.
Tambiah, Stanley J. 1997. Leveling Crowds: Ethnonationalist Conflicts and Collective Violence in
South Asia. University of California Press.
Tec, Nechama. 1987. When Light Pierced the Darkness: Christian Rescue of Jews in Nazi-occupied
Poland. Oxford, UK: Oxford University Press.
Tilly, Charles. 2003. The Politics of Collective Violence. Cambridge University Press.
Tilly, Charles. 2008. Contentious Performances. Cambridge University Press.
Turner, Ralph H. and Samuel J. Surace. 1956. “Zoot-Suiters and Mexicans: Symbols in Crowd
Behavior.” American Journal of Sociology 62(1):14–20.
Ulfelder, Jay and Benjamin Valentino. 2008. “Assessing Risks of State-Sponsored Mass Killing.”.
Varshney, Ashutosh. 2002. Ethnic Conflict and Civic Life: Hindus and Muslims in India. Yale
University Press.
Varshney, Ashutosh. 2007. Ethnicity and Ethnic Conflict. In Oxford Handbook of Comparative
Politics, ed. Carles Boix and Susan Stokes. Oxford, UK: Oxford University Press pp. 274–294.
Voigtl¨ander, Nico and Hans-Joachim Voth. 2012. “Persecution Perpetuated: The Medieval Origins
of Anti-Semitic Violence in Nazi Germany.” The Quarterly Journal of Economics 127(3):1339–
Volkov, Shulamit. 2006. Germans, Jews, and Antisemites: Trials in Emancipation. Cambridge,
UK: Cambridge University Press.
Welch, David. 2002. The Third Reich: Politics and Propaganda. New York, NY: Routledge.
Wildt, Michael. 2014. Hitler’s Volksgemeinschaft and the Dynamics of Racial Exclusion: Violence
Against Jews in Provincial Germany, 1919 - 1939. Berghahn Books.
Wilkinson, Steven I. 2004. Votes and Violence: Electoral Competition and Ethnic Riots in India.
Cambridge University Press.
Wood, Elisabeth Jean. 2009. “Armed Groups and Sexual Violence: When Is Wartime Rape Rare?”
Politics & Society 37:131–161.
Yanagizawa-Drott, David. 2014. “Propaganda and Conflict: Evidence from the Rwandan Geno-
cide.” The Quarterly Journal of Economics 129(4):1947–1994.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Historians are increasingly studying lynching outside of the American Southeast, but sociologists have been slow to follow. We introduce a new public data set that extends existing data on lynching victims to cover the contiguous United States from 1883 to 1941. These data confirm that lynching was a heterogeneous practice across the United States. We differentiate between three different regimes over this period: a Wild West regime, characterized mostly by the lynching of whites in areas with weak state penetration; a slavery regime, found in former slave states, characterized mostly by the lynching of blacks; and a third minor regime, characterized by the lynching of Mexican nationals mostly along the Texas-Mexico border. We also note great variability at the county level in the extent of lynching. By contrast, we find very little state-level variability in lynching once local and regional regimes are considered. We discuss the implications of local and regional heterogeneity for quantitative lynching research using these data.
Cambridge Core - Political Sociology - Protectors of Pluralism - by Robert Braun
" Democracy and Displacement in Colombia's Civil War offers novel and important research on how and why violence is deployed during civil wars. "-Winifred Tate, author of Drugs, Thugs and Diplomats. " Democracy and Displacement in Colombia's Civil War has all of the hallmarks of a classic work on forced migration and Colombian politics. It is a pleasure to read, well argued, and carefully researched. "-Idean Salehyan, author of Rebels without Borders Democracy and Displacement in Colombia's Civil War is one of few books available in English to provide an overview of the Colombian civil war and drug war. Abbey Steele draws on her own original field research as well as on Colombian scholars' work in Spanish to provide an expansive view of the country's political conflicts. Steele shows how political reforms in the context of Colombia's ongoing civil war produced unexpected, dramatic consequences: democratic elections revealed Colombian citizens' political loyalties and allowed counterinsurgent armed groups to implement political cleansing against civilians perceived as loyal to insurgents. Combining evidence collected from remote archives, more than two hundred interviews, and quantitative data from the government's displacement registry, Steele connects Colombia's political development and the course of its civil war to purposeful displacement. By introducing the concepts of collective targeting and political cleansing, Steele extends what we already know about patterns of ethnic cleansing to cases where expulsion of civilians from their communities is based on nonethnic traits.
Why do pogroms occur in some localities and not in others? Jeffrey S. Kopstein and Jason Wittenberg examine a particularly brutal wave of violence that occurred across hundreds of predominantly Polish and Ukrainian communities in the aftermath of the Nazi invasion of the Soviet Union. The authors note that while some communities erupted in anti-Jewish violence, most others remained quiescent. In fact, fewer than 10 percent of communities saw pogroms in 1941, and most ordinary gentiles never attacked Jews. Intimate Violence is a novel social-scientific explanation of ethnic violence and the Holocaust. It locates the roots of violence in efforts to maintain Polish and Ukrainian dominance rather than in anti-Semitic hatred or revenge for communism. In doing so, it cuts through painful debates about relative victimhood that are driven more by metaphysical beliefs in Jewish culpability than empirical evidence of perpetrators and victims. Pogroms, they conclude, were difficult to start, and local conditions in most places prevented their outbreak despite a general anti-Semitism and the collapse of the central state. Kopstein and Wittenberg shed new light on the sources of mass ethnic violence and the ways in which such gruesome acts might be avoided.
Cambridge Core - International Relations and International Organisations - Resilient Communities - by Jana Krause