0
$\begingroup$

I have been looking for a method similar to quantiles but using means instead medians, but I couldn't find anything.

I mean a procedure where we obtain the mean of the whole sample $\bar{X}$. Then take the mean of only the data above $\bar{X}$, and next the mean of only the data below $\bar{X}$. We get here the mean and two 'submeans' around the mean. We don't need to stop here and can further continue subdividing the sample if there is enough data.

For instance for the sample {0.167, 0.177, 0.181, 0.181, 0.182, 0.183, 0.184, 0.186, 0.187, 0.189}, the mean is 0.1817; the 'submeans' are 0.1765 and 0.1852; the 'subsubmeans' are 0.1670, 0.1797, 0.1830, and 0.1873. And so on until finally we obtain the original values of the sample in the fifth iteration.

I don't know if this method exists, and if it exists I don't know the name and cannot find information about it.

Improve this question
$\endgroup$
2
  • 2
    $\begingroup$ They are vaguely reminiscent of trimmed means although they are usually trimmed symmetrically. $\endgroup$
    – mdewey
    Commented Sep 5, 2017 at 14:18
  • 1
    $\begingroup$ @mdewey Yes, the submeans would be like trimming only from one side. $\endgroup$
    – juanrga
    Commented Sep 5, 2017 at 15:24

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.

Browse other questions tagged or ask your own question.