Measuring Elusive Populations with Bayesian Model Averaging for Multiple Systems Estimation: A Case Study on Lethal Violations in Casanare, 1998-2007

Abstract

Collecting data for the analysis of past human rights violations is fraught with challenges. For example, individuals from or about whom data should be collected may be displaced, missing, or dead. Some reports of acts may be easier to find than others, and as a result, datasets will be biased toward those cases. These challenges must be overcome in order to create effective official policy for violence mitigation and prevention. This relies on statistical analyses that can meet these challenges.We propose a combination of Bayesian Model Averaging and Multiple Systems Estimation as an example of this type of analysis, which presents an important advancement in human rights research. In particular, this method allows the use of multiple data sources to estimate the number of undocumented violations -those that have not been recorded by any source.We present an application of this method in Casanare, Colombia, where we estimate a total (both documented and undocumented) of 5,832 killings (95% credible interval: 3,822, 9,332) and 2,345 disappearances (1,221, 4,901) between 1998 and 2007.

Civilian killings and disappearances during civil war in El Salvador (1980‒1992)

“…Researchers also frequently use MSE to estimate the size of under-registered, stigmatized populations, such as users of illegal drugs (e.g., Mastro et al 1994), unhoused people (e.g., Berry 2007), and sex workers (e.g., Geibel et al 2007). MSE has been applied to conflict mortality in several cases (for an overview, see Brunborg, Tabeau, and Urdal 2006;Ball and Price 2019), including Guatemala (Ball et al 1996), Kosovo (Ball et al 2002), Peru (Ball et al 2003), Bosnia (Tabeau 2009), and Colombia (Lum et al 2010). More recently, MSE has been used to estimate the number of people held in slavery (e.g., Bales, Hesketh, and Silverman 2015, and a November 2017 special issue of Chance magazine).…”

Bayesian Population Size Estimation Using Dirichlet Process Mixtures

“…Lum et al. () analyzed multi‐list data documenting deaths due to political violence in the Casanare department, Colombia. This dataset has also being analyzed by Mitchell et al.…”

“…Using a Bayesian model averaging approach, considering only three lists at a time, Lum et al. () estimated (95% interval ) killings.…”

“…This means that the BIC log-linear estimates ignore the sampling variability inherent to the data-based model selection. In contrast, the NPLCM model simultaneously estimates all the parameters Lum et al (2010) analyzed multi-list data documenting n = 2629 deaths due to political violence in the Casanare department, Colombia. This dataset has also being analyzed by Mitchell et al (2013), and a previous version by Guzmán et al (2007).…”

“…The first real dataset consists of four incomplete overlapping lists documenting killings during the Kosovo war (Ball et al., ). The second is a collection of 15 matched lists documenting killings due to political violence in Casanare, Colombia (Lum et al., ). In the Supplemental Materials online, we include an additional example using a classical small‐sample dataset from the ecological literature.…”