Do Evalues have any value?
evalue.pdf (73.4 KB)
It’s essentially another way of thinking about effect sizes in observational studies. E-values per se don’t convey additional information that is independent of the original effect size estimate reported, they just give the reader another way of restating the results.
It’s a bit similar to the fragility index and P-values. The former is basically just a restatement of the latter, but some people find it helpful to rephrase the original quantity (P-value; a probability) in a different way (fragility index; number of events).
As such, there’re probably not many scenarios where the 2 would lead to widely diverging results. If you have a very high E-value, that means you have an impressively large effect size (spurious or not). If you have a very large fragility index, that means you have a very low P-value.
2 Likes