While time series data has advanced in bounds in both quantity and quality “beyond the dreams of intellectual avarice,” due to the painstaking archival work of economic historians, careful archaeological eldwork, and technological progress in paleoclimatic modeling, the statistical tools utilized in data analysis have thus far trailed behind the data in many applied settings in anthropology, archaeology, and economic history. In the context of Medieval England, Campbell (2010) advanced the idea that ecological variation was a major driver of historical human population dynamics, via the positive and negative feedback mechanisms linking climatic variation, crop yields, crop prices, and human population growth. Likewise, Kennett et al. (2012) utilized climatological reconstructions, to argue that drought and consequential political balkanization was a key driver of the collapse of Mayan polities. Zhang et al. (2007) argue that historical fluctuations in temperature on the macro-scale are associated with war, population declines, and agricultural price inflation. However, across all of these studies, and many others like them, analyses were limited mostly to basic descriptive statistics and visual comparisons of key events in time series data. While these method of data analysis provide an initial characterization of structure in the data, more formal methods of time series analysis, especially when situated in a Bayesian framework, provide a principled way in which patterns not easily visible in the raw data can be investigated.
Paper coming soon.