[jph00]

The SK death rate of 0.7% assumes that all 98% of the cases with outcome currently classified as "unknown" will recover. That isn't rational! You need to do proper survival analysis to account for the growth in cases.

Here's a brief analysis of the current numbers from SK: https://twitter.com/jeremyphoward/status/1237324005108924417

[kaiabwpdjqn]

I think the real takeaway is from all this is that death rate is a pointless metric. It is highly dependent on the local demographics, it requires precise information which is rarely available, it is biased by the level of care available, it has numerous ways to estimate it all of which are hard to explain and not actual estimates but upper or lower bounds, it tends to naturally decrease over time, etc.

The death rate is lower in China than Italy, but the death rate is lower within each age group in Italy vs China. Most people are too innumerate to understand this statement.

As an aside, I REALLY hate how this guy on twitter says “it could be 5.0%, look at this spreadsheet that assumes 5.0%!” Then refers to a paper as a good analysis which claims 1.6% and a set of facts which differ greatly from all of his assumptions.