I’ve been exposed to a record number of papers making basic errors with hypothesis tests. On my bike ride to work I started thinking about what it would take for me to not want to enrol myself in a cohort (we have a few important cohorts here at Erasmus MC). It led me down the following line of thought.
People become outraged when they see things like p=0.06 and the authors treat their results as though they’re significant. But no one gets equally upset if p=0.06 and people use that to say that the null is true even when the confidence intervals are very wide and include a lot of important values. That is just as egregious, so why do people get way more upset with one error and not the other?
