• Following is a message from one of my graduate students. Any suggestions for alternative methods to compare management zones for significant differences would
Message 1 of 2 , Jul 5, 1999
View Source
• 0 Attachment
Following is a message from one of my graduate students. Any
suggestions
for alternative methods to compare management zones for
significant
differences would be appreciated.

Steve Searcy
----------------------------------------------
Hi,

I'm a graduate student at Texas A&M working in the precision farm
area through the ag engineering department. My advisor, Steve
Searcy, suggested that write to this group and ask for suggestions.

I have broken a 100 acre field into 20 management units based on
some criteria. I have very intense yield data. On average, there
are 3000 yield points within one management unit. I'm trying to
prove that the management units that are adjacent are different
from each other (based on yield). The statistical tests that I run tell
me that my results are not significant (the adjacent management
units are
not different); however, when I graph the results I can see
differences in
average yields. I also tried a variety of other methods to determine
management units, and I'm getting the same results.

I am running a Student's t-test. The difference in the yield
average between the adjacent management zones
ranges from 3 to 20 and I have very high variances. Within the
approximately 3000 points that I have in one management unit, the
yield
ranges from 8 to 238 which accounts for the high variance (300-
400). Due
to the high variance and small difference between the yield
averages, the
t value is very small (around .1 to .01). When I compare this value
to a
Bonferroni t value (95% confidence) it is very insignificant (the
Bonferroni t value is around 2.4).

I get the impression that these types of statistical tests have been a
problem with analyzing precision farming techniques. I was
wondering if
anyone had any suggestions for me concerning other statistical
tests or
other ways to prove differences with this kind of data.

Sincerely,
Jennifer Jessip
AGEN-Research Assistant

*************************************************************************
Stephen W. Searcy E-mail: s-searcy@...
104
AERL Phone: 409-845-3668 West
Campus
Rt. 4A Fax: 409-8478627 Texas A&M University
College Station, TX 77843

--
*To post a message to the list, send it to ai-geostats@....
*As a general service to list users, please remember to post a summary
of any useful responses to your questions.
*To unsubscribe, send email to majordomo@... with no subject and
"unsubscribe ai-geostats" in the message body.
DO NOT SEND Subscribe/Unsubscribe requests to the list!
• Jennifer, You have your answer in front of you. The average yields ARE noticeably different. You can see the differences in graphs and other methods. Whether
Message 2 of 2 , Jul 5, 1999
View Source
• 0 Attachment
Jennifer,
You have your answer in front of you. The average yields ARE noticeably
different. You can see the differences in graphs and other methods.
Whether or not the 3 to 20 yield differences are meaningful is something
you'll have to decide. However the variances are large, even when based
on 3000 samples. This is very common in environmental data. It means
either that the natural variability is inherently large or that you have
not controlled enough extraneous variance to detect a statistically
significant difference.

So what do you do? Here are some options.

1. Look into an analysis of variance (or covariance, depending on your
variables) design that will control additional variance. If you have data
that can be used in an ANCOVA design, you might be able to detect the
differences.

2. Just report what you found. Show the graphs and other analyses. But
instead of the statistical tests, report confidence limits. They convey
the same information without the stigma of not being significant.

Charlie Kufs

~~~~~ TERRABYTE CORPORATION
~~~~~ Statistics and Computer Technologies
~~~~~ Applied to the Environmental Sciences
~~~~~ http://members.home.net/terrabyte
~~~~~ terrabyte@...

On Mon, 5 Jul 1999 09:22:44 -0600 "Gary Sallows"
<SallowsG@...> writes:
>Following is a message from one of my graduate students. Any
>suggestions
>for alternative methods to compare management zones for
>significant
>differences would be appreciated.
>
>Steve Searcy
>----------------------------------------------
>Hi,
>
>I'm a graduate student at Texas A&M working in the precision farm
>area through the ag engineering department. My advisor, Steve
>Searcy, suggested that write to this group and ask for suggestions.
>
>I have broken a 100 acre field into 20 management units based on
>some criteria. I have very intense yield data. On average, there
>are 3000 yield points within one management unit. I'm trying to
>prove that the management units that are adjacent are different
>from each other (based on yield). The statistical tests that I run
>tell
>me that my results are not significant (the adjacent management
>units are
>not different); however, when I graph the results I can see
>differences in
>average yields. I also tried a variety of other methods to determine
>management units, and I'm getting the same results.
>
>I am running a Student's t-test. The difference in the yield
>average between the adjacent management zones
>ranges from 3 to 20 and I have very high variances. Within the
>approximately 3000 points that I have in one management unit, the
>yield
>ranges from 8 to 238 which accounts for the high variance (300-
>400). Due
>to the high variance and small difference between the yield
>averages, the
>t value is very small (around .1 to .01). When I compare this value
>to a
>Bonferroni t value (95% confidence) it is very insignificant (the
>Bonferroni t value is around 2.4).
>
>I get the impression that these types of statistical tests have been a
>problem with analyzing precision farming techniques. I was
>wondering if
>anyone had any suggestions for me concerning other statistical
>tests or
>other ways to prove differences with this kind of data.
>
>Sincerely,
>Jennifer Jessip
>AGEN-Research Assistant
>
>
>*************************************************************************
>Stephen W. Searcy E-mail: s-searcy@...
>104
>AERL Phone: 409-845-3668 West
>Campus
>Rt. 4A Fax: 409-8478627 Texas A&M University
>College Station, TX 77843
>
>
>--
>*To post a message to the list, send it to ai-geostats@....
>*As a general service to list users, please remember to post a summary
>of any useful responses to your questions.
>*To unsubscribe, send email to majordomo@... with no subject
>and
>"unsubscribe ai-geostats" in the message body.
>DO NOT SEND Subscribe/Unsubscribe requests to the list!

___________________________________________________________________
Get the Internet just the way you want it.
Free software, free e-mail, and free Internet access for a month!
Try Juno Web: http://dl.www.juno.com/dynoget/tagj.
--
*To post a message to the list, send it to ai-geostats@....
*As a general service to list users, please remember to post a summary
of any useful responses to your questions.
*To unsubscribe, send email to majordomo@... with no subject and
"unsubscribe ai-geostats" in the message body.
DO NOT SEND Subscribe/Unsubscribe requests to the list!
Your message has been successfully submitted and would be delivered to recipients shortly.