Accounting for uncertainty in conflict mortality estimation: An application to the Gaza War in 2023-2024

The Gaza War, triggered by the Hamas-led attack on October 7, 2023, has resulted in significant loss of life and intensified an ongoing humanitarian crisis. Despite increasing demand for accurate measures of conflict severity, mortality estimates remain challenging due to the inherent `statistical fog of war' surrounding ongoing conflicts. In particular, accurate quantification is hindered by uncertainties related to incomplete reporting and uncertain age-sex distributions of casualties. Official death tolls are likely influenced by damaged infrastructure, security disruptions, and political motivations, as well as the aggregated nature of available data. This complicates detailed demographic verification. Our study introduces a novel methodological approach---a Bayesian model incorporating novel priors---to explicitly account for measurement errors in mortality estimation by addressing reporting completeness and uncertainty in demographic distributions. Applying these methods to the ongoing Gaza War, we estimate sex- and age-specific mortality patterns and associated life expectancy (LE) losses. We find that LE in Gaza was 42.3 [39.4-45.0] in 2023 and 40.4 [37.5-43.0] in 2024, corresponding to LE losses of 34.4 [31.7-37.3] and 36.4 [33.8-39.3], respectively. To contextualize these estimates, we compare them with LE losses observed in the Gaza Strip, the West Bank, and all of Palestine between 2012 and 2019. Our estimates align with point estimates of previously published work, after adjusting the reporting priors to ignore underreporting. The study provides a robust and reproducible framework for mortality estimation under conditions of data scarcity, applicable to current and future conflicts.