Naked Statistics: Stripping the Dread from the Data

by Charles Wheelan

For that reason, I have adopted a common convention, which is to take the higher standard error of the two and use that for all of the candidates.