Quantitative Social Science and the Holocaust
History is replete with violence, which casts a long shadow on contemporary attitudes, behavior, and institutions, as covered in our earlier posts. The Holocaust is the largest single episode of mass violence in the 20th century. It is also one of the most “data-rich” genocides, and the number of books, articles, and films that deal with it is enormous.
Despite the abundance of records and testimonials, quantitative social scientists have been slow to engage with this tragedy, leaving it to philosophers, social theorists, and historians. In 2012, Charles King observed that the event’s uniqueness, the amount of suffering, and the centrality of memory in Holocaust discourse made it a challenging subject for political scientists who are interested in generalization and comparison.
Yet quantitative social scientists no longer shy away from the subject. In the last five years, four articles that used the Holocaust to understand capacity for sustained armed resistance, the legacies of repressive institutions, the consequences of wartime property transfers, and the willingness to protect persecuted minorities appeared in the APSR, a flagship political science journal. Top academic presses published books about the role of state power, religious pluralism, experience with repression, and political polarization in the survival of Jews as well as on the politics of remembrance in post-war Europe. In January 2020, Jeffrey Kopstein, Jelena Subotic, and Susan Welch convened historians, political scientists, and sociologists for a conference the University of California Irvine and an edited volume on the social science of the Holocaust.
This post discusses what we have learned from just one sliver of recent quantitative work on the Nazi extermination and concentration camps. This work generated a lot of discussion and ignited some controversy, but carries important lessons for historical political economy research more broadly. While produced by political scientists, it draws on rich literature in other disciplines, which painstakingly reconstructed the design of the camps, the lists of victims, the train routes, the economy within camps, and the relations between camp prisoners, camp guards, and the surrounding communities.
First, some background. Between 1933 and 1942, Nazi Germany established a network of extermination and concentration camps across Central and Eastern Europe (see map). Dachau camp was the first and served as a model for other concentration camps on the German territory. These early camps housed not only Jews, but also German Communists, Socialists, Freemasons, Jehovah’s Witnesses, Roma and Sinti, homosexuals, repeat criminal offenders, and other political and ideological opponents of National Socialism. The death rates in concentration camps were staggering, but these camps were not explicitly designated for extermination. In 1941-42, camps explicitly for the purpose of killing were built in occupied Poland. Over 2.7 million people were murdered in Kulmhof, Belzec, Sobibor, Treblinka, Majdanek and Auschwitz-Birkenau with poison gas or by shooting. Concentration camp labor, as well as the property seized from the victims of concentration and extermination camps, enabled and financed the German war effort.
How did this man-made geography of repression shape political attitudes and behavior in postwar Europe? Did witnessing and benefitting from mass violence have long-run consequences for the surrounding communities? Recent quantitative work is a testament to the durability of the violent past, but leaves many issues unresolved.
In a 2017 article, Evgeny Finkel and I examine how wartime property transfers affect political attitudes and behavior at the community level. We focus on the Treblinka death camp because its location is exogenous to local social and economic conditions. The camp was built in 1942 in the secluded rural area; no Jews lived in the surrounding countryside before the war and a separate railroad was constructed to transport the victims to the camp from the main railway juncture. The local Polish population did not work at the camp, and the camp’s operations were initially kept secret. After the war, Treblinka experienced few commemorative activities; a small on-site museum opened as late as 2006. These factors help us to isolate the impact of property redistribution from other aspects of conflict, including preexisting intergroup relations in affected communities, the upheaval that accompanies violence, and subsequent remembrance.
We argue that the main first-order impact of the camp was economic: the local population traded with the camp guards and prisoners; some later dug through the ashes to uncover valuables of the deceased. We show that distance to the camp predicts the share of houses constructed in the immediate postwar period and the quality of roof material in the 1980s. We also find that proximity to the camp had enduring political consequences: in the 2001 election, when the Holocaust was salient, support for the far-right anti-Semitic party was higher in the vicinity of the camp. We propose – but do not directly test – two channels to explain the political legacy of Treblinka: material interest and cognitive dissonance. The beneficiaries of Jewish property acquired a stake in the new status quo; Jewish claims to restitution became a threat to their welfare, however remote. Benefitting from the misfortune of others also generates cognitive dissonance, which can be reduced by vilifying the victims and denying wrongdoing. Both channels feed support for political parties that deny ingroup wrongdoing, derogate the victims, and oppose restitution.
The subsequent work also uses distance to the site of mass violence as an explanatory variable, but expands the selection of sites to all concentration camps Germany and emphasizes different mechanisms. In a 2019 article, Julian Hoerner, Alexander Jaax, and Toni Rodon find a positive effect of proximity to a concentration camp between 1933 and 1945 on support for far-right parties in the 2013 and 2017 federal elections. The authors view camps as physical monuments and attribute the findings to a “memory satiation effect.” They argue that German voters near the former camps confront the past more frequently, which makes them “more receptive to revisionist historical accounts.” Thus, commemoration of the Holocaust that emphasizes ingroup guilt can backfire, increasing support for political parties that challenge the narrative about ingroup wrongdoing. The authors find null results in the former East Germany, where the “contrition frame” was less salient historically. The study thus raises questions about how historical legacies interact with and are transformed by subsequent policies.
In a 2020 article, Jonathan Homola, Miguel Pereira, and Margit Tavits examine how proximity to the concentration camps in Germany affects attitudes and behavior using data from the European Values Survey and the 2017 federal elections. They find that Germans who live closer to the former camps are more likely to display outgroup intolerance and vote for the far right. The authors contend that confronting the ideology of the Nazi regime leads to cognitive dissonance that is reduced by adopting a negative view on ethnic minorities.
These three articles use the same explanatory variable, proximity to a former camp, and find largely consistent patterns: support for the far right increases with proximity to the site of mass violence. This is striking because they focus on populations with dramatically different war and postwar experiences (Poles vs. Germans). At the same time, the authors rely on different causal pathways to explain their findings and cannot directly test the theorized mechanisms. In addition, the sites of mass violence they study are not randomly distributed and outcomes are measured more than half a century after the war, during which many observed and unobserved post-treatment variables can affect the outcome.
Evgeny Finkel and I limited our study to Treblinka because other Nazi camps were located near cities with large Jewish population or in the regions that experienced mass displacement after the war. When research design cannot address endogeneity issues directly, considerable additional work is necessary to establish confidence in the results. Concentration camps were often established in places with appropriate infrastructure or needs for forced labor; such locations developed differently both before and after the war. For example, Flossenbürg, Mauthausen, and Natzweiler-Struthof were located next to granite deposits, some of which would be used for construction of the infamous Nazi Party Rally Grounds in Nuremberg. Some camps, such as Dachau, were subsequently used to house German expellees after the war, which led to competition over victimization discourses and could plausibly increase both support for the far right and outgroup intolerance.
It is instructive to look at the recent debate over the Homola et al. (2020) study to understand these issues. In a replication of this article, Tom Pepinsky, Sara Wallace Goodman and Conrad Ziller attributed some of the findings to unobserved spatial heterogeneity. They show that adding fixed effects for contemporary German states reduces or even reverses the coefficients in the original paper. Homola et al. respond to this criticism in this article. They note that including fixed effects reduces statistical power and increases sample variance, which makes this empirical approach suboptimal for analyzing surveys that sample respondents in clusters from a limited set of locations. Their main counterargument, though, is that including fixed effects for contemporary federal structure introduces post-treatment bias. Homola et al. show that the original results – greater intolerance in the vicinity of the camp – hold when pre-WWII states are used for estimating fixed effects or when contemporary states are specified as post-treatment in sequential g-estimation analysis.
Another, arguably more important issue that is not covered in this debate but is essential to interpretation and generalizability of the findings is the paltry evidence on the mechanisms. All three papers cannot test the individual-level psychological theories they build on. Two of the three articles measure outcomes using aggregate electoral data, and surveys used in Homola et. al. (2020) does not allow measuring cognitive dissonance.
In order to examine the mechanisms, additional work is necessary. Researchers could manipulate messages about the past to evoke varying degrees of cognitive dissonance to evaluate whether different narratives about ingroup complicity in mass violence evoke different responses. The memory saturation effect can be tested by comparing cohorts that received different doses of Holocaust education or by looking at attitudes toward outgroups before and after a commemorative event. Scholars can also compare former camps that experienced different “fates” after the war to back out the mechanisms, though these divergent postwar trajectories may be endogenous to local preferences.
We should also take the implications of intervening post-treatment events more seriously, particularly when dealing with electoral behavior. It is by now a classic move of Europe’s far right to emphasize suffering of the ingroup and deny ingroup culpability in the suffering of the outgroup, including by questioning facts about the Holocaust and denying their nations’ wrongdoings in WWII. Holocaust denial and distortion are so prevalent on social media that last week Facebook and Twitter announced the ban on such content. Without these counter-frames, living next to the former concentration of extermination camp may be conducive to intergroup reconciliation. For example, in an online experiment in Poland, I find that narratives about ingroup wrongdoing reduce ethnocentric beliefs, but when the perpetrator narrative is countered with a story about ingroup victimhood, ethnocentrism remains high and support for minority rights is diminished.
The Holocaust has shaped how we think about the legacies of violence, the politics of race and ethnicity, the dangers of dictatorship and populism, and the responsibility of governments to accept refugees and provide humanitarian assistance. It remains important for social scientists and policymakers today, against the background of rising prejudice against minorities and the spread of disinformation and conspiracy theories on the web.