Gene Ballay: Pitfalls to consider with COVID stats
Regardless of where we fall on the pandemic opinion spectrum, we all need to have at least a basic understanding of the pitfalls associated with the statistics that are being quoted.
We each want what’s best for the country, but as various authorities request, even demand, that the public accept their recommended treatment protocol, there are some fundamental data issues which they need to address.
First is the question of independent validation: there is a long-term, on-going controversy surrounding the validity of multiple cornerstone medical research papers.
Response time may dictate that replication-additional study take place at a later date, but those of us who came of age during the 1950s and 1960s remember well the thalidomide scandal, purportedly one of the greatest man-made medical disasters in history.
With regard to “statistical success,” there is a long history of improper study conclusions. Assuming the mistake has been made innocently (not profit based), the mis-characterization often happens when the focus is upon observed success, when performed on a sample set that is designed to be “positive,” while not recognizing failed results that were observed on the sample set that was known to be “negative” — The Bayesian Trap.
Authorities may soon request (demand) yet more stringent public responses, but no one (media or otherwise) is discussing basic data issues.