Lies, Gosh Darn Lies, and Not Enough Good Statistics: Why Epidemic Model Parameter Estimation Fails

Daniel E Platt; Laxmi E Parida; Pierre Zalloua

If, as tracking numbers increased, testing surveillance was broad enough to pick up community spread individuals proportional to total numbers of tests applied, then the proportion of positives from the tests may reflect population rates. However, if rates are tightly limited to immediate known cases, then the reported positives will be a better estimate of underlying population, since the fraction of those seeking medical assistance should be proportional to the exposed number in the population. When available tests increased, the apparent rate grew substantially. Therefore, infected population growth may be more closely reflected in the fraction of positive results normalized by total number of tests applied, in spite of very highly biased sampling selection. For a given proportion of ill patients who seek help, this should track . CC-BY-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.