In recent years, social scientists have started taking on board the idea that cultural norms, attitudes, and beliefs are important determinants of behavior. A growing body of work theorizes about their origins, considers their relationship to institutions, and explores their significance for important questions in economics and political science.
Since I am a historian, and historians these days focus primarily on questions of culture (broadly speaking), this should be welcome news to me. But I’m actually a bit skeptical about this turn to culture and beliefs. One of the things that has always appealed to me about a social science view of the world is that it provides a way of understanding human behavior without resorting to culture or “mentalities”. It’s not that I think these things don’t matter; they obviously do. But how do they matter? My skepticism is about our ability to isolate, define, and measure culture – particularly for societies in the past. Specifying the role of culture with any precision is difficult enough in a contemporary context where we have a much better intuitive grasp of the qualitative context of the societies we live in, and where we can ask people about their views and what they think they are doing in any given situation. (Though there is still the obstacle that they – we – are not consciously aware of many aspects of their – or our – own cultural embeddedness.) Historical societies, however, leave only scattered hints about their belief systems in their cultural context. We often have no sense of what we might be missing. Historians usually have to triangulate among widely disparate kinds of sources in order to obtain some inkling about cultures and beliefs, and even then their conclusions remain highly tentative.
This is especially true when it comes to broad generalizations about the beliefs of ordinary people. So many of the sources we have for the past were generated by literate elites and their views were not necessarily those of the other 98 percent. The voices of ordinary people are not often heard in archival sources. The information we have about their lives is usually extracted from sources that were designed for quite different purposes: parish registers, household inventories, wills, land and tax registers, marriage contracts and dowries, dispute resolution records (court documents and depositions), or petitions (for poor relief, for settlement rights, for permission to marry). These sources provide us mainly with information about what people did, and not what they thought. And the sources that do record the voices of ordinary people – police records, court records, petitions – are not without ambiguities. Is the argument made by the plaintiff based on her beliefs? Is she invoking universally accepted norms or rhetoric she knows will appeal to the court? Do the views portrayed in these sources reflect the culture of the larger society or the culture of the governing elite or that of some ill-documented subculture? How might we find out? (An interesting example of this problem is discussed in this book, which describes what happened when it was decreed that judges in rural Russian courts should decide peasant disputes in accordance with – unwritten – customary law.)
Even the most meticulously researched accounts of norms and beliefs run into problems with ambiguity in the interpretation of findings and the inherent difficulty of isolating beliefs as causal factors. An example that comes to mind is a fascinating book I read recently on marriage patterns in imperial Russia. (Yep, marriage patterns again!) In Russian Peasant Women Who Refused to Marry: Spasovite Old Believers in the 18th-19th Centuries, John Bushnell investigates the remarkably high female celibacy rates on serf estates in imperial Russia. Using a range of sources in creative ways, Bushnell links these high female celibacy rates to sectarian beliefs – in particular, the views of Spasovites, an Old Believer sect whose members thought they would be damned if they accepted sacraments (such as marriage) from Orthodox priests. There is no ‘smoking gun’ here – the voices of the unmarried women themselves are not heard in the sources and their reasons for not marrying are never explicitly noted. However, the evidence from other sources – petitions from people complaining to authorities about the deviant behavior of non-marrying women, notes about the presence of sectarian communities in specific regions, documents related to investigations of schismatics – bolsters the correlation and makes the argument plausible.
But what is especially interesting – and especially confounding for the argument that marriage behavior was belief-driven – is the variation. The serf estates where female celibacy ran between 10 and 40 percent were overwhelmingly in the Central Industrial Region, where local economies were more diverse and women had greater opportunities for employment outside agriculture. Even more interesting is that high female celibacy rates varied across estates governed by the same landlord (where sectarians were present), depending on whether the estates were located in the agricultural (low female celibacy) or proto-industrial (high female celibacy) zones. To complicate things further, the marriage behavior of male sectarians was not deviant in the same way. It was female Spasovites who chose to remain unmarried in specific contexts.
That beliefs might interact with material conditions on the ground to generate heterogeneous outcomes is hardly shocking – especially in the context of demographic behavior. It’s the question of how to interpret the role of beliefs that remains ambiguous here. Was it the case that the economic opportunities available to women in a wage work/quitrent economy enabled them to act in accordance with their sectarian beliefs, where the more coercive system of agrarian serfdom in the Black Earth region did not? Or could it be that, in areas where women could support themselves, Spasovite beliefs offered ideological cover for opting out of a social arrangement that was in many ways more advantageous to men and put women at increased risk of domestic violence and maternal mortality? (Why did Spasovite men not avail themselves of the same opportunities?) Or were Spasovite beliefs in fact quite irrelevant to the marital choices of these women? How can we make that determination when the group in question left us nothing to go on but marital status, which happens to correlate with a wide range of other variables?
The difficulties of separating culture and beliefs from concrete incentives that derive from institutions and material conditions are well known. Both Jared and Vicky have discussed this problem in recent blog posts. I do not mean to imply here that I think these difficulties make the study of culture and beliefs in the past pointless. Not at all – these are interesting and important questions. Again, my skepticism is about measurement and causality. Most studies of culture tend to raise more questions than they answer, precisely because of these ambiguities of interpretation. If a good institution is hard to find, a good proxy for a particular belief is even harder. Any proxy for cultural attributes inevitably has multiple interpretations. What does even something as apparently straightforward as “proportion Catholic” tell us? Is it measuring people’s genuine, devout belief in the tenets of the Church? Is it measuring sociability? The value of community? The costs of deviation from the norm? Some aspect of the environment we can’t see in our sources? Does one know this historical society well enough to control for all other correlates with Catholicism? For all other possible explanations of the observed behavior? Many of the ambiguities of interpretation arise from problems related to defining culture or cultural attributes. Can things like “individualism” or “collectivism” or “trust” be specified precisely enough to make clear they must have been cultural artifacts and not responses to some overlooked or unobservable constraint in the environment? Not surprisingly, the methodological challenges posed by such questions are addressed in a large literature that spans a range of disciplines, including biology, cognitive science, anthropology, and economics.
In this attempt to spell out my concerns, I’m reminded of one of those disciplinary differences that Scott noted in his post on historians and social scientists (and in our interdisciplinary conversation): the notion of “sources” versus that of “data”. Historians do worry about sources – in ways that we often don’t articulate very convincingly. In these interdisciplinary contexts, our questions and comments tend to be heard by our social science colleagues as different versions of “you can’t use those numbers for that” with few compelling arguments as to why not. The importance of sources is so obvious to historians, so internalized to the practice of our discipline – we aren’t used to being challenged on this question. Sources are the very foundation of historical research: they shape the questions we ask – they determine which questions we can ask – and the context in which we ask them. They are the source of our knowledge about the society in question, and at the same time they tell us what we cannot know about this society. The silences are as important as the information on the page. Given this, I think what historians are often wondering when they ask about sources is “do you know how this source/these data fit(s) into the larger picture?” or “how much do you understand about the society you are studying?”. In some cases, these questions aren’t really relevant to the research and come across as unnecessarily critical, but in the case of culture, norms, beliefs, or attitudes, they matter a great deal.
Empirical social science research sheds important light on questions of culture and norms when it brings concrete evidence of observed behavior to bear against conventional cultural explanations. In the case of imperial Russia, the prevailing view of peasant society in the historiography is one of conformity and overwhelming social control. Russian peasant society was, in this view, collectivist, patriarchal, and allowed limited scope for female agency. The evidence uncovered by Bushnell’s study undermines this view of a monolithic rural society. It indicates a robust culture of subversion among (some) peasant women, as well as among the male family members who supported their choices. And it seems that this form of deviance was widespread in places where the local political-economic framework left room for it. This raises a number of interesting questions about historical political economy and institutions at the local level and their effects on people’s behavior – questions that our sources and methodologies equip us to address much more convincingly.