1
$\begingroup$

I am looking at a test strategy for Covid-19, where I want to combine two tests with different sensitivity and specificity.

Visualizing this would probably make sense in a "1-sensitivity" vs "1-specificity" graph. Is there a name for these quantities?

Informally (in particular in my head) I talk about false negative vs false positive rate, but I have already realized this is ambigous, since people will have different intuitions what I am normalizing to.

$\endgroup$
4
$\begingroup$

You can refer to $1-$ Sensitivity as the "false negative rate" (aka $\beta$ error or Type II error), and $1-$ Specificity as the "false positive rate" (aka $\alpha$ error or Type I error).

You can see it on the right-side box on the Wikipedia entry: https://en.wikipedia.org/wiki/Sensitivity_and_specificity

$\endgroup$
3
  • $\begingroup$ I like the wording, but from experience it is unclear what the denominator is: 1) all people tested or 2) all who were tested positive (ie people confuse with 1-NPV) or 3) all who are truly positive (1- sensitivity, what we want). $\endgroup$ – Jonas Heidelberg Dec 13 '20 at 8:42
  • $\begingroup$ But of course with the Wikipedia as a reference that will clear it up for most people ;-). I might actually start using „miss rate“ and „fallout“, which are given there as synonyms. Have you seen those used IRL? $\endgroup$ – Jonas Heidelberg Dec 13 '20 at 8:45
  • 2
    $\begingroup$ I've seen "false positive rate" and "false negative rate" used. I haven't seen "fallout" used, and I wouldn't really know what it refers to. "miss rate" I think is a bit clearer, but I still think you'd need to explain what you mean when you use it because it's not common. $\endgroup$ – Santiago Romero Brufau Dec 13 '20 at 16:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.