jmp-l
[Top] [All Lists]

Re: jmp incorrectly calculates mean

To: jmp-l@lists.biostat.wustl.edu
Subject: Re: jmp incorrectly calculates mean
From: Ronald Cross <roncross@cox.net>
Date: Wed, 28 Dec 2005 17:26:12 -0800
In-reply-to: <20051228224640.22000.qmail@web30806.mail.mud.yahoo.com>
References: <20051228224640.22000.qmail@web30806.mail.mud.yahoo.com>
I would love to but the $500.00 plus price is steeped. I have found what is causing the problem and I have a work around for this. It appears to be occurring whenever I have a very large std dev in the data column for a given level. For example, most of the data will be between 1.4 and 1.8 but then there are a few outliers that will read 200000000 or more. If I go to range check and adjust the values to something reasonable like 1.2 to 2.0, then the mean is correctly calculated. The disadvantage of course is that the outlier is no longer available for investigation: it turns into a missing value.

Any advice on how I can get a cheaper upgrade version of JMP for Macs. I would be willing to pay no more than $200.00. I asked my company about buying a few Mac licenses, but they shot back and asked why am I using a Mac, they stated, "you should be using what everyone else is using." So if I am going to upgrade, I will have to purchase it for myself. I am lucky that they didn't make me throw my machine away.

thx
RLC

On Dec 28, 2005, at 2:46 PM, chem engineer wrote:

Ron - you should move up to a newer version ...

Ronald Cross <roncross@cox.net> wrote: Hi all, I am using JMP 4.0.4 for mac running a G4 powerbook with
1GByte of RAM with a 1GHz power pc processor on OS X 10.4.3.
However, this version of jmp runs under OS 9 (classic).
I have noticed that when I do a fit y by x plot and display the
mean and standard deviation, it incorrectly calculate the mean of
each level where y is continuous, numeric and x in nominal. It
appears to be a round off error of some sort. For example, when I
use table summary on the same data, the mean will be roughly 1.4, but
in the mean and std dev display of the fit y by x plot, it appears as
a 1 and have a std dev of 0. Std dev is 0 for all levels since all
numbers are rounded off to the same whole number.

I have loo k in the column info section of the data column and best is
selected as the format. It doesn't occur for all columns and I am
not sure why I am seeing it on some columns and not others.

Any ideas on solving this would be appreciated.

thx
RLC


Yahoo! Shopping
Find Great Deals on Holiday Gifts at Yahoo! Shopping


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