The affect of outliers.

azsakura · #1 $Old$ December 5th, 2008, 02:29 AM

i have 2 set of scores.
1. 213,322,262,242,159,100,80,59,23,52,89,215
2. 30,30,29,27,25,23,22,23,25,28,29,30

in the first set, there are 2 outliers, 322 and 23.
i know that with a high outlier it drags the score up, with a low outlier it pulls it down. How can i explain the 2 extremes occuring at the same time?

and the range will be larger so that it can not represent the data range correctly. Is this correct?

thank you.

CaptainBlack · #2 $Old$ December 6th, 2008, 02:34 AM

Quote:

Originally Posted by azsakura $View Post$

i have 2 set of scores.
1. 213,322,262,242,159,100,80,59,23,52,89,215
2. 30,30,29,27,25,23,22,23,25,28,29,30

in the first set, there are 2 outliers, 322 and 23.
i know that with a high outlier it drags the score up, with a low outlier it pulls it down. How can i explain the 2 extremes occuring at the same time?

What do you mean by "explain the 2 extremes occuring at the same time?"?

Is there anything to explain? Recompute the statistics with each of them removed, and with both of them removed.

Quote:

and the range will be larger so that it can not represent the data range correctly. Is this correct?

Depends on the origin or the outliers, if they are legitimate data from the tails of the sampling distribution then that is one thing, if they are due to experimental error and are not legitimate samples then discard them.

CB