0
$\begingroup$

If it is of any one's interest, the data is taken from the duration of men "last in bed" as studied here.

I want to calculate the percentiles (say, 95th) of a non-normal distribution without raw data. What are the minimum constraints needed? I have the following and I think some are not necessary.

  • n=474
  • mean=8.5
  • SD=7.47
  • median=6.0
  • min=0.1
  • max=52.7
| cite | improve this question | | | | |
$\endgroup$
  • $\begingroup$ There's an infinite number of possible solutions. It might be possible to derive some bounds $\endgroup$ – Glen_b -Reinstate Monica Apr 1 '19 at 14:46
1
$\begingroup$

This is not possible. The percentile depends on the exact shape of the distribution. So unless you make a parametric assumption about the distribution or an assumption directly about the 95th percentile you cannot estimate the percentile well without the original data.

| cite | improve this answer | | | | |
$\endgroup$

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.