0
$\begingroup$

I’ve been reading some papers in the neuroscience field, and I don’t quite understand the widespread use of AUC/ROC to test for group differences when analyzing neuronal firing over a range of seconds (with regular intervals between observations).

I mean, this approach does capture differences in signal/effect magnitude between groups. But if you apply an inferential model to compare only the area under the curve instead of the raw signal, wouldn’t that underestimate the errors in the model, since you’re only using the "systematic part" of the observation? On top of that, doesn’t the use of AUC/ROC ignore the shape of the effect?
What are the limitations of using AUC/ROC in inferential analyses of time-varying signals?

Improve this question
$\endgroup$
1
  • $\begingroup$ Welcome to Cross Validated! This page (among others on this sie) has extensive discussion on AUROC. I think you'll find that discussion to validate at least some of your concerns. After reading that page, come back to edit this question to be specific about what still isn't clear. It might help also to add some web links to examples of the use of AUROC for time-varying signals in neuroscience that trouble you. $\endgroup$ Commented Sep 24 at 19:17

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.