0
$\begingroup$

Lets say we have a set of arbitrary numbers, whose values range from $1$ to $100$

$$ x = \{x_1,x_2,x_3,x_4,x_5,...,x_n\} $$

Given only the common summary statistics of $x$ (mean, standard deviation, variance), but not to $x$ itself.

Would it be possible to calculate or even estimate the trimmed mean (the mean excluding the lower and upper 10% of outliers) to a reasonable extent?

$\endgroup$
0
$\begingroup$

No, imagine 2 distributions, one a symmetric bell curve, and the other a spike centered at 50 with asymmetric outliers around 0 and 100. Their averages could both be 50, their standard deviations and variance could be the same, and their max and min would be the same, and their medians would be the same.

However, removing the outliers from the spiky distribution will change the mean, but the same is not true for the bell curve

see also https://stats.stackexchange.com/questions/135737/will-two-distributions-with-identical-5-number-summaries-always-have-the-same-sh

$\endgroup$

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.