Conversation

I like the value of the S-value. But after 3 attempts to use the S-value in a manuscript, its use by co-authors or editors/reviewers was rejected on the grounds that was unusual and too confusing to use. #epitwitter

@Lester_Domes

Quote Tweet

Our R package, which produces confidence (compatibility/consonance) intervals at every level (95%, 90%, 80%, 40%...) and plots them to form a function, is now on CRAN. You can take a look here. data.lesslikely.com/concurve/ @avigotsky

Show this thread

Replying to

You could report S-values only and a table showing the correspondence between S-values and p-values.

4:31 PM · Jun 25, 2019Twitter for Android

Replying to

Here's what we did in a recent paper, we provide the P-value, corresponding S-value, maximum-likelihood ratio, and the likelihood-ratio statistic to show the relationships. Details of computations can be found in the appendix arxiv.org/abs/1909.08579

Replying to

Yes, we did, in fact, use the word "quintupling" and we have no regrets