# Decision Identifiability

Seasonal influenza models are routinely asked to recommend an antiviral coverage tier from weekly reported cases, yet a well-calibrated parameter posterior does not make that tier identifiable. A decision-relevance score weighted by the variance of the decision margin measures how much of that variance comes from directions the data failed to shrink; on a 192-cell synthetic grid the bounded posterior still flipped the plug-in action in 26.7% of cells, and the plug-in rule lost 11.7% of the oracle's value. On 204 real FluView region-seasons, widening the reporting-rate prior raised the main arm from 0.476 to 0.733 while the scale-invariant control stayed near zero.

Seasonal influenza models are routinely asked to recommend an antiviral coverage tier from weekly reported cases, yet a well-calibrated parameter posterior does not make that tier identifiable. A decision-relevance score weighted by the variance of the decision margin measures how much of that variance comes from directions the data failed to shrink; on a 192-cell synthetic grid the bounded posterior still flipped the plug-in action in 26.7% of cells, and the plug-in rule lost 11.7% of the oracle's value. On 204 real FluView region-seasons, widening the reporting-rate prior raised the main arm from 0.476 to 0.733 while the scale-invariant control stayed near zero.