Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


Math Forum
»
Discussions
»
Courses
»
apstat
Notice: We are no longer accepting new posts, but the forums will continue to be readable.
Topic:
Re[2]: Minitab's outliers
Replies:
0




Re[2]: Minitab's outliers
Posted:
Sep 25, 1996 2:00 PM


Peter, First, thanks for your reply. This has me really wondering about how the guys out in the world use the terms. I thought I understood what was the common usage, and now I am totally unsure. I would like to at least know that I'm preparing my students with language appropriate for the AP exam. I guess I can see a logic to your definition of median as a range of values for some situations but,I thought the definition was fairly consistant, at least I'm satisfied that the way I teach it my students will be able to communicate with other people in the field successfully. I do think it is a little confusing if you apply the range of values idea to percentiles . If I were to tell you I interviewed three people and 50% of them were male, I think your first reaction would be to look at me somewhat suspiciously. I feel much the same discomfort in saying that the 50th percentile of (0,0,1) was 0. It is true that 50% of the ordered data were at or below this value, and 50% were at or above this value, but then I could say the same thing for any value from 34 to 66. I guess I'm more concerned about having my students able to understand what is meant when it is written in common statistical usage but I would like (hope) for that usage to be as common and mathematically reasonable as possible. Pat Ballew Misawa, Japan
______________________________ Reply Separator _________________________________ Subject: Re: Minitab's outliers Author: Peter Blaskiewicz <pjb@mcc.cc.tx.us> at EDUINTERNET Date: 9/24/96 12:20 PM
One of my esteemed statistics teachers in college several years ago defined the median for us as a value that is chosen/calculated in such a way that 1) at least 50% of the ordered data was at or below that value, and 2) at least 50% of the ordered data was at or above that value. He extended this definition in a natural way to cover percentiles also. Note that, under this definition, it is not necessary that there be a unique median for a given set of data. For example, in the data set {0,1}, every number in the closed interval from 0 to 1 would qualify as a median. However, when we accepted this as our working definition of median, it helped us resolve many problems. Peter Blaskiewicz



