Chapter 4 A Novel Solution to Biased Data in Disaster Management: COVID-19 Example

Abstract Disaster management often faces missing data replaced by biased approximations. For example, a lack of randomised testing over the general population during the unprecedented COVID-19 pandemic created biased data regarding its incidence. This chapter describes the problem and its solution from the Econometrics literature, unbeknownst to epidemiologists and public health professionals. Our model has a choice equation and an outcome equation to overcome the bias measured by the ‘inverse Mill’s ratio’ (IMR). We illustrate using state-level sampling to create a large sample of data and report forecasts of COVID-19 deaths. Since using the IMR reduces bias and improves out-of-sample forecasts, it has broader applicability in disaster management. Disasters like droughts, floods, and earthquakes have socio-economic consequences involving choice probabilities by human agents.