Proximity Ascertainment Bias in Early Covid Case Locations
arXiv:2401.08680 · doi:10.1093/jrsssa/qnae021
Abstract
A comparison of the distances to the Huanan Seafood Market of early Covid cases with known links to the market versus cases without known links shows results apparently incompatible with a location model lacking proximity ascertainment bias. The sign of the difference instead agrees with a model in which such ascertainment bias is large. In the presence of such bias inferences based on the clustering of case locations become unreliable.
short, v2 updates a reference and notes that linked and unlinked cases are not displaced in the same direction from the market. v3 more refs. v4-small correction to an angle, remembered that latitude and longitude are not equivalent except at equator. v5 includes response to Debarre/Worobey Reply