We present evidence of a false discovery rate over 90% for medical observational studies. We show how this high false discovery rate can occur. The operational paradigm for the analysis of medical (and environmental) observational studies is "no correction for multiple testing" and "no sharing of data sets."
Some multiple testing mistakes are due to ignorance, but others are intentional, following a well-articulated (faulty) scientific paradigm; over $1B of grant/tax money flows to institutions with reproducibility problems revolving around a multiple testing and other statistical issues, e.g. model selection.
Return to Biostatistics Working Group