Rule of Thumb for Sensitivity and Specificity What are rule of thumb cutoffs for sensitivity and specificity? What is high, good, fair, low?
 A: Amplifying @AdamO's comment, there are two sorts of issues here:


*

*How high can both of them be?

*How high should each of them be, relative to the other?


The first question will depend on the state of the field you are working in. There are some things (diseases or whatever, I will use "disease" from here on, but it could be something else) where we have very good methods of detection. There are other diseases where we have nothing very good at all. Perhaps the disease has multiple etiologies; perhaps the diagnostic variables are only loosely related to the disease; perhaps the diagnostic methods are not very accurate or reliable; perhaps the connections between the diagnostic variables and the disease are different in different groups in ways that we don't yet understand.
The second question depends on the relative cost and benefit of different sorts of errors and correct decisions. There are four:


*

*True positive. If the disease can be treated easily and has a good cure rate, this is very high. If there is no treatment at all, then a true positive may have very little value.

*True negative. This may help avoid expensive, painful or risky treatment.

*False positive. Can cause psychological stress, incorrect treatment (which may have risks) and so on.

*False negative. The disease progresses.
So, what disease are you diagnosing? Cataracts or colon cancer?
You will have to read up on the literature in your field to answer either question and you may have to do some do some deep thinking to answer the second.
