The comment above is partially true. The data sources for each country are different, and thus have different error profiles. It's statistically incorrect to

The comment above is partially true. The data sources for each country are different, and thus have different error profiles. It's statistically incorrect to assume they can be merged and treated as a single data set. What is more useful is to examine well bounded and vetted subsets of the data. Once the South Korean cases become more mature in a couple weeks, I think we will have very high confidence mortality data. Including poor data quality sources will just introduce biases.