Measuring the Impact of Armed Conflict on Population Health: A Guide for Researchers.

The humanitarian impact of armed conflict remains a significant international issue, with an estimated 2 billion people residing in fragile or conflict-affected settings. Despite increasing attention and study of armed conflict and its impact on human populations, few studies have evaluated the methods necessary to assess such relationships: how to use disaggregated and granular conflict data, measure and operationalize conflict, etc. In this study, we identify important considerations for conducting armed conflict and health research, including how data structures and decisions might impact conclusions. We discuss the particular characteristics of existing armed conflict datasets and the types of biases that may be present in data drawn from conflict-affected areas. Further, we demonstrate how data and measurement choices can result in different conclusions and, if handled improperly, even spatial misclassification, bias, and spurious conclusions. Lastly, we illustrate some of these data and measurement choices in an empirical example where we assess the impact of conflict on women's contraceptive use in Nigeria. Using conflict event data from the Uppsala Conflict Data Program and health data from the Nigerian Demographic and Health Survey in 2018, we show how the relationship between exposure and outcomes varies across different spatiotemporal dimensions of conflict exposure measurement.