Raw Data on Covid-19 Positive Tests and Deaths at the State Level
Raw data on Covid-19 has been characterized by cyclical reporting irregularities, especially underreporting on certain days of the week and overreporting on others. It’s plagued with significant data classification errors, which tend to get corrected in large batches. For example, many states combined data on virus tests with data on antibody tests. When they decided to separated the data for the two kinds of tests, they reported negative numbers of tests for the day. There have been numerous instances of states deciding that more people probably died from Covid-19, which has resulted in as many as 1500 deaths or more being added on one day. Some states have identified people whose deaths had been attributed to Covid-19 who didn’t really die from Covid-19, and that has resulted in states occasionally reporting negative numbers of deaths on certain days.
For all these reason, I find the raw Covid-19 data to be of limited usefulness. If you review the graphs below, you will see just how irregular the state-level reporting has been.