Towertalk
[Top] [All Lists]

[TowerTalk] FW: more accurate statistically

To: "towertalk" <towertalk@contesting.com>
Subject: [TowerTalk] FW: more accurate statistically
From: "W5GN" <w5gn@mxg.com>
Date: Tue, 11 Aug 2015 06:57:56 -0500
List-post: <towertalk@contesting.com">mailto:towertalk@contesting.com>
The average human has one tit and one ball.

So much for an average being a useful descriptive metric

I concur TOTALLY on percentile (median) metrics. 
It's what REAL Operating Systems on REAL computers 
(z/OS on IBM z/13 MainFrames) use for resource management.
We tell z/OS we want 95 % of tranactions in 1 second, for
a particular class of work, as the goal, and z/OS manages
resources (moving real memory and/or real engines) to 
meet that goal.

And no matter HOW LARGE a sample set, a bimodal distribution
can still have an average value that is not a value in the
sample set.

73

Barry, W5GN


Herbert W. "Barry" Merrill, PhD
President-Programmer
MXG Software
Merrill Consultants
10717 Cromwell Drive
Dallas, TX 75229-5112
Barry@MXG.COM
Fax:  214 350 3694 - Still works, received as email
Tel:  214 351 1966 - Unreliable, please use email

www.mxg.com        HomePage: FAQ answers most questions
admin@mxg.com      License Forms, Invoice, Payment, ftp information
support@mxg.com    Technical Issues 
MXG-L FREE ListServer  http://www.mxg.com/mxg-l_listserver/



-----Original Message-----
From: TowerTalk [mailto:towertalk-bounces@contesting.com] On Behalf Of
kr2q@optimum.net
Sent: Monday, August 10, 2015 6:04 AM
To: towertalk@contesting.com
Cc: towertalk@contesting.com
Subject: Re: [TowerTalk] more accurate statistically

K8RI replied:
[snip]
That is true for small samples, but not those large enough to be
statistically significant.
The larger the sample, the more accurate statistically. 
[end snip]

Roger:
This (above) is a totally different statement from what you originally
stated and to which I commented.  

I don't care how big the sample population is, the fact is that the mean
(average) will always be more sensitive to outlier data when compared to
applying the median to the same set of data.

Determining which measure of central tendency is "more accurate" or "more
meaningful," for whatever you may want to learn about, is a completely
different topic.

de Doug KR2Q
_______________________________________________



_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk

_______________________________________________



_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk

<Prev in Thread] Current Thread [Next in Thread>