When comparing statistical tests, one often compares

  1. the probability of committing a Type 1 error, and
  2. the probability of not committing a Type 2 error.

While (2) can be concisely termed 'power', is there a concise term for the (1)?

(I don't think significance is the right term)

  • 1
    $\begingroup$ See en.wikipedia.org/wiki/…. $\endgroup$ – whuber Dec 31 '14 at 0:10
  • 3
    $\begingroup$ The radar literature calls the two probabilities as the "false-alarm probability" and "false-dismissal probability" which terms I find easier to remember than the bland and unimaginative "Probability of Type 1 error" and "Probability of Type 2 error". $\endgroup$ – Dilip Sarwate Dec 31 '14 at 0:16
  • $\begingroup$ @whuber, I suppose by that link you are suggesting 'false positive rate'? But just like I prefer 'power' over 'sensitivity' because it sounds more generic (or discipline-independent), I still wonder if there is an equivalent term for (1). But thanks for the link anyway. $\endgroup$ – qoheleth Dec 31 '14 at 1:21
  • $\begingroup$ Specificity is the only one-word term I can think of. I don't find it memorable, so I just say "$1-\alpha$". Nobody gets confused about what that means. $\endgroup$ – Glen_b -Reinstate Monica Dec 31 '14 at 2:03
  • $\begingroup$ A nominal size. $\endgroup$ – Khashaa Dec 31 '14 at 6:04

The pithiest word I've seen is size

  1. Significance level $\alpha$, see here
  2. Power $1-\beta$, sensitivity or recall rate
  • $\begingroup$ as mentioned in my original question, I don't think 'significance' is the right term. Yes, $\alpha$ certainly coincides with the numerical value of (1), but I think it is describing the other side of the same coin. After all, there is nothing significant about rejecting a correct null hypothesis. $\endgroup$ – qoheleth Jan 5 '15 at 7:00

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.