A Bayesian Approach to Survivorship Bias in Historical Data Analysis [Ein Bayes'scher Ansatz zum Survivorship Bias in historischer Datenanalyse]

Wand, Tobias

Abstract

Datasets such as Seshat have allowed researchers to quantitatively test hypotheses about premodern societies and states with great success. Nevertheless, one has to consider potential sources of bias in the data such as a survivorship bias favouring the inclusion of long-lived over short-lived states. Bayesian methods can be used to complement standard modelling procedures to take this issue into account as is demonstrated by analysing the longevity distribution of premodern states.