How Can Statistical Methods of Social Sciences Complement Historical Research?
Seeing faces of individual victims of the Holocaust, such as those of the prisoners of the Theresienstadt ghetto displayed by the publicly accessible database developed by the Theresienstadt Initiative Institute, is a critical element of remembering and understanding the Holocaust. Such databases often provide a near-complete coverage of a particular Holocaust setting: a camp or a population of a city, with many characteristics of victims available to the analyst. Hence, such databases offer the potential of exploring the Holocaust using statistical analysis. When can statistical methods that seek to recover central tendencies, as opposed to trying to understand individual histories, be a useful complement of historical research of the Holocaust? First, multivariate statistical methods allow one to condition on several characteristics at once when describing (quantifying) the relationship of a given factor to coping strategies of victims or to their success in avoiding persecution and ultimately death.1
Second, in some instances, statistical methods allow one to zoom in on a situation that involves quasi-random assignment of victims to varying conditions and environments, thus allowing for quantification of causal effects of such assignments. In social sciences, such methods have been used for decades to understand social phenomena, from the labor market to social behavior, and they can be useful in the study of the Holocaust as well, as I discuss, together with Tomáš Jelínek, in a recent exploratory study.
Social Status and Social Links
“Mutual help of prisoners in concentration camps was the main and most effective way of saving lives.”
Testimony of an Auschwitz survivor2
This testimony is one example among many in a large literature predominantly based on Holocaust survivor accounts. That small groups providing mutual support are critical for survival in extremely harsh conditions has been proposed in the study of not only Nazi concentration camps, but also of POW camps and Soviet Gulags.3 However, since much of the existing literature is based on survivor testimonies, and since these are fundamentally selective, it is plausible that one is applying an inappropriate interpretation to these testimonies. It is possible that those who did not survive also formed mutual-support groups and that such groups were therefore not critical for survival. This possibility makes statistical analysis based on all prisoners of a given internment setting an important complement of qualitative research. Ideally, one would like to observe existing social linkages (friendships) among all prisoners and ask whether prisoners who were more isolated in the social structure of a ghetto or camp were less likely to survive. A social scientist would additionally seek to study a situation where being isolated is not driven by one’s unobservable characteristics (such as health status) that might drive survival in their own right, but when being isolated was determined by a chance — a quasi-random assignment. Such comparisons would speak to the causal nature of social ties in surviving the Holocaust.
While there is little quantitative research on social linkages in the Holocaust, there are several recent studies on the degree to which pre-war social structures (including one’s social status) are reflected in the internal operation of societies within Nazi camps and ghettos.4 We ask whether pre-war social status and both pre-war and imprisonment-based social ties are important determinant of who survives the extreme conditions of Nazi camps and ghettos. To understand the role of social status, one would ideally like to systematically compare its effects on prisoners’ outcomes across varying settings, under different organization of prisoner societies, and under different survival pressure dictated by external conditions. Were different types of social resources helpful to prisoners facing different types of survival pressure?
In a recent study, joint with Tomáš Jelínek and Matěj Bělín, we rely on the near-complete database of Theresienstadt prisoners in an attempt to make headway on these questions. In absence of direct information on prisoner friendships, we employ social-linkage proxies (potential friends) based on various pre-existing networks. We then statistically examine the importance of pre-deportation social status and of the availability of social networks (linkages, potential friends) for survival in two Holocaust settings: Theresienstadt and Auschwitz. More specifically, we study the entire Jewish population of the Theresienstadt ghetto5 and separately those Theresienstadt prisoners who ended up on transports to Auschwitz-Birkenau. Since our analysis is based on the near-complete database of individual histories of Theresienstadt prisoners, which records deaths in Theresienstadt as well as ultimate Holocaust survival for all prisoners of the ghetto, it avoids survival biases by incorporating information on those who did not survive the Holocaust.
Theresienstadt prisoners faced high levels of stress and omnipresent hunger even before entering the extreme environment of the Auschwitz-Birkenau labor and extermination camp. Pre-existing friendships, social and family ties may be particularly valuable in such settings, where there are few market substitutes for social resources. For each prisoner in Theresienstadt, we construct a variety of proxies (listed below) for an individual’s social status, and for the availability of potential friends. We do the same for prisoners on transports out of the ghetto, as of their arrival at Auschwitz.6 We then ask how these two types of social-capital measures affect major risks faced by prisoners: (i) selection into deadly transports out of Theresienstadt, (ii) death in Theresienstadt, and (iii) death after transport to the Auschwitz-Birkenau concentration camp.
In the first step of the analysis, we estimate statistical models focused on prisoner selection into transports out of Theresienstadt. The ultimate destination of transports to the East was not disclosed to the self-administration. We condition on the externally determined (SS-specified) demographic composition of transports and study the relative risk (for prisoners in the at-risk demographic groups) of being on the next transport.7 This relative risk inside one’s demographic group was affected to a significant extent by decisions taken by the Jewish self-administration of the ghetto, which was in charge of selecting individual prisoners onto transports out of Theresienstadt. Prisoners isolated in the social space of the ghetto and those of low social status may have found it hard to seek patronage or employment in the internal administration. We therefore assess the importance for this selection into transports of an individual’s social status (for example, having the status of a ‘prominent’ prisoner or having been a member of a pre-war business elite group), and of membership of Jewish pre-deportation self-administrations (Jüdische Kultusgemeinde in Prague, Israelitische Kultusgemeinde in Vienna, Berlin; henceforth referred to as JKG/IKG), which were strongly involved in the ghetto’s self-administration. We study these questions across the varying degrees of group-level selection risk given by SS decisions, and find that social status and JKG/IKG membership protect against selection, particularly so during periods of high group-level selection risk, i.e., particularly when a large portion of a prisoner group was to be selected for the next transport.
Next, we quantify determinants of the risk of dying in Theresienstadt using a death hazard (duration) model, which measures how strongly prisoners’ characteristics, the degree of overcrowding, and the timing of arrival in Theresienstadt predict dying in the ghetto. We focus on older prisoners for whom the probability of dying in Theresienstadt was high, and contrast death risk structures across varying levels of ghetto-wide survival pressure. We estimate what appears to be the first multivariate death hazard model from Nazi concentration camps/ghettos (these types of models are widely used in health and social sciences). We confirm that isolated elderly prisoners are particularly vulnerable to ghetto overcrowding, hunger, and disease. In addition to family- and social-linkage effects, we again find that social status and membership of pre-deportation self-administrations have protective effects. These effects are again stronger when prisoners face higher ghetto-wide death risks, i.e., during periods when the probability of dying in the ghetto was high.
Finally, we study Holocaust survival among the 17 thousand Theresienstadt prisoners who entered Auschwitz-Birkenau, a large labor and extermination camp. We focus on quasi-random differences in the social-linkage composition of transports to Auschwitz8 in order to contrast survival chances across prisoners depending on whether they entered the camp alongside a group of socially-linked potential friends, or as socially isolated prisoners who may find it difficult to form friendships (reciprocal relationships) in the camp.9 In this extreme context, we consider as ‘socially linked’ prisoners who: had family ties on the transport; shared a place of residence prior to deportation to Theresienstadt10; were formerly interned together in a non-deadly agricultural-labor camp11; were linked by a distribution chain of an underground satirical weekly (‘Shalom on Friday’) while in Theresienstadt; arrived in the Theresienstadt ghetto on the same in-transport; or had JKG/IKG connections.
We find a significant survival advantage conferred by entering Auschwitz with several socially-linked fellow prisoners, based on measures reflecting potential close-friendship links and camaraderie, but not based on social linkage proxies corresponding to pre-deportation administrative ties and social status. In addition, we find that prisoners who were particularly apt at avoiding selection into transports out of Theresienstadt before eventually being deported to Auschwitz were also more likely to survive there. This finding connects an indirect realized proxy of social standing within Theresienstadt with survival chances in Auschwitz, and provides further support for the notion of social-capital effects.
By studying the survival value of social linkages across different environments and varying survival pressure, we extend the existing empirical literature on the Theresienstadt ghetto. By contrasting Theresienstadt to Auschwitz, we compare factors that underpin survival across qualitatively different survival pressure and also across prisoner societies with and without the ability to self-administer. We thus shed light on the structure of a prisoner society and social resources underpinning survival in extremity. We also provide a novel quantification of the survival effects of prisoners’ academic titles, including medical degrees, in addition to that of basic demographic characteristics. Our results, based on the most systematic statistical exploration of individual histories available to-date from a Nazi concentration camp setting, confirm the findings of qualitative work based on selective survival testimonies that being socially isolated was particularly costly during the Holocaust.
Our analysis generates complementary evidence to, and a statistical check on the large part of the Holocaust literature based on fundamentally selective survival testimonies. Given the selective nature of survivor testimonies used in qualitative research, our findings help cement their interpretation and lessen their methodological criticism based on survival bias arguments. Our analysis supports the qualitative literature in its emphasis on the importance of mutual-support groups as a key survival strategy of prisoners facing extreme survival pressure and offers novel quantitative findings as well: Within the self-administered society of the Theresienstadt ghetto, pre-deportation social status and administrative linkages helped protect prisoners against selection into out-transports and supported survival of the elderly. But social status and administrative linkages were no longer helpful in the extreme conditions of the Auschwitz-Birkenau concentration camp, where friendships corresponding to shared previous residence, earlier shared imprisonment, and prisoner networks all generated a significant survival advantage.
Our evidence can be extended and checked by study of specific episodes, testimonies and other documents. For example, we identify individuals who were statistical outliers in their ability to avoid deportation out of Theresienstadt, after we control for all available factors, including measures of social status and social linkages. Since this ability to avoid many transport selections in Theresienstadt also helps predict survival after transport to Auschwitz, it is likely that it is related to prisoners’ characteristics that we do not observe in the data available to us. Historical research can ask whether there is a common factor, not available to us, connecting these prisoners and providing a potentially new explanation for their survival ability.
Our evidence is also relevant to the literature studying parochial altruism—the notion that experience of violent conflict supports within-group cooperation among survivors.12 An alternative mechanism highlighted here is that those more prone to cooperation (thanks to, e.g., having available larger social networks) are more likely to survive violent conflicts. Finally, our analysis extends the literature on the importance of social links in high-stakes contexts13 and the large literature on the importance of social networks for health outcomes14 by providing evidence on the transferability of social linkages generated in normal social environments to the truly extreme conditions of deadly internment camps.
- For example, researchers of the Holocaust sometimes compare survival rates across nationality groups or across cohorts; such univariate comparisons can be made more precise by simultaneously controlling for several other observed factors that affect the outcome of interest. ↩
- Tomáš Radil, Holocaust a Evropa po sedmdesáti letech (Holocaust and Europe after Seventy Years), (Prague: Academia Press, 2016, 165. ↩
- Shamai Davidson, “Human Reciprocity Among The Jewish Prisoners in the Nazi Concentration Camps”, in Yisrael Gutman and Avital Saf (eds.), The Nazi Concentration Camps – Structure and Aims, the Image of the Prisoner, the Jews in the Camps (Jerusalem: Yad Vashem, 1984), 555–572; John McElroy, This Was Andersonville (New York: McDowell, Obolensky, 1957); Paul Schmolling, (1984): “Human Reactions to the Nazi Concentration Camps: A Summing Up”, Journal of Human Stress, 10 (1984): 108–120; Anne Applebaum, Gulag: A History (New York: Doubleday, 2003). ↩
- For example, Evgeny Finkel, Ordinary Jews (Princeton: Princeton University Press, 2017); Maja Suderland, Inside Concentration Camps (Cambridge, UK: Polity Press, 2013). ↩
- The Theresienstadt ghetto was established in 1941 by the SS in German-occupied Czech lands. Most of the prisoners had been deported from Czech, German, and Austrian cities. Of the 140 thousand Theresienstadt prisoners, 33 thousand died there, almost all elderly, and over 80 thousand were sent to extermination camps. See pages 60-62 of Jurajda and Jelínek (2021) for a more detailed description of the ghetto. ↩
- We do not study prisoners deported to Treblinka and Maly Trostinec, the other two chief destinations of out-transports from Theresienstadt, as virtually none of these survived the Holocaust. Unlike Auschwitz, Treblinka and Maly Trostinec had no labor camp component, they were solely extermination camps, and so social status of links could be of no help to prisoners entering these camps. ↩
- In other words, we ask whether, within a prisoner demographic group that was facing a given group-level risk of being selected for the next transport, say 10%, certain individual prisoners (for example those with stronger social ties or status) fared better relative to the 10% group-wide average risk of selection. ↩
- In the analysis of prisoners on transports to Auschwitz-Birkenau we explore variation in social-linkage resources based on the sorting of prisoners into transports: in some transports of a given size, there were larger groups of potential friends than in others. We consider the composition of out-transports from Theresienstadt in terms of social linkages a ‘laboratory’ for our study of the effect of social networks on survival. While the evidence discussed earlier shows prisoners with higher social status and resources were less likely to be selected into transports to the East, most prisoners eventually ended up on transports due to the immense transport pressure, and there is no evidence that the composition of out-transports was optimized in terms of social linkages available on transports. ↩
- We thus avoid not only the selection bias of survival testimonies, but also potential omitted variable bias where prisoners who are more pro-social or healthier tend to have more friends and are more likely to survive. ↩
- We observe street addresses prior to deportation for the large group of former Prague residents, and also town of residence for several small Czech towns. ↩
- Testimonies of prisoners of the agricultural camp (in the Czech town of Lípa) suggest that social ties formed in this previous imprisonment were helpful in extermination camps. (Stránský and Ullmann, 1990) ↩
- Robert L. Trivers, “The Evolution of Reciprocal Altruism,” The Quarterly Review of Biology 46 (1971), 35–57; Jung-Kyoo Choi and Samule Bowles, “The coevolution of parochial altruism and war,” Science, 318 (2007), 636–40. ↩
- Dora E. Costa and Matthew E. Kahn, “Surviving Andersonville: The Benefits of Social Networks in POW Camps,” The American Economic Review 97 (2007), 1467–1487. ↩
- James S. House, Karl R. Landis and Debra Umberson, “Social relationships and health”, Science 241 (1988), 540–545. ↩
