Why can't we keep using p-values and also have an addendum to papers that includes the raw data observed and their associated likelihood functions?