details many of the on-farm tests performed by farmers with help
from fieldmen, extension agents, and researchers in the Pacific
Northwest during 1993-94. Most are connected in some way to the
STEEP II On-farm Testing project, a federally funded effort aimed
at helping farmers make use of on-farm tests to increase development
and adoption of soil and resource conserving farming methods. The
OFT project is lead by extension workers from Idaho, Oregon, and
Washington, and provides technical assistance and educational materials
to farmers through their local Cooperative Extension office. We
also work with Ag industry, Conservation Districts, and other organizations.
For more information on the project or on-farm testing, contact
Stewart Wuest (509-335-3491). Also see the article on designing
on-farm tests and the guide to resources later in this bulletin.
One of the
most important aspects of a well designed on-farm test is replication.
This is why I have put the data from each replication in the reports.
Statistics help in deciding if measured differences in performance
are due to the treatments applied or simply due to the random variation
that exists in every field. If the difference between the treatments
is greater than the "least significant difference at a 5% probability
level" (LSD(5%)), we know that there is only a 5% probability
that it was caused by random variation and not the treatments. Personal
judgement and thought must go into any conclusion, and final decisions
regarding a farming practice should reflect many economic and management
factors and more than one year of tests. On-farm tests are designed
to provide farmers with an accurate, low risk tool for exploring
production options and making successful decisions.
an On-farm Test
test design presented here is aimed at measuring and comparing the
performance of two or three different crop management strategies,
or "treatments". An on-farm test can be used to measure
performance in terms of yield, stand establishment, protein, water
infiltration, weed counts, or other criteria. The treatments could
be as modest as the application of fungicide in one treatment and
not in the other, or as different as zero-till seeding compared
to plow, cultivate, and seed, or even a comparison of different
test that will produce accurate, conclusive information requires
replicated, side-by-side comparisons. This is the only way to distinguish
yield differences that occur naturally between two strips from differences
actually caused by the treatments. Extensive research in the Inland
Pacific Northwest has shown that long, narrow, side-by-side strips
replicated four to six times can produce a very accurate comparison.
The longer the strips are, the better the data is likely to be.
There have been many successful tests with four replications of
300 ft strips, but 750 ft or longer strips are more likely to produce
accurate results. Four replications are recommended, but five or
six replications should be used if the comparison might produce
small or very important differences. It is difficult to understand
the importance of adequate replication until you have had some experience
trying to draw conclusions from data with only two or three replications.
Try to resist the temptation to minimize the number of replications.
With the availability of portable weighing equipment, eight to twelve
strips can be harvested in less than three hours.
what the treatments are going to be, pick locations in the field
where you can place the treatments in long, side-by-side strips.
All strips in a replication should have an equal chance to perform
well, in your best judgement. In other words, do not place one strip
on flat ground and the other on a hill slope. Other areas to avoid
are fence lines and field corners where extra fertilizer and tillage
occur. Flip a coin to decide which treatment goes in which strip.
Repeat for each replication. Replications can be next to each other,
or in separate parts of the field.
are made, such as stand counts or yield, record them separately
for each strip. The data can be analyzed statistically using a hand
calculator and step-by-step formulas, a free, easy-to-use computer
program from OSU called AGSTATS, or with help from your county extension
agent. Even without statistics, a lot can be learned by looking
at each replication to see if one treatment was consistently better
than the other.
If you are
a beginner at doing experiments, ask for some help from your extension
agent or someone with OFT experience. Most likely a little discussion
with an experienced experimenter will save a mistake or two and
make your OFT more successful.
Northwest On-farm Test Results. S.B. Wuest, B.C. Miller, R.J.
Veseth, S.O. Guy, D.J. Wysocki and R.S. Karow. 1992. Department
of Crop and Soil Sciences Technical Report 92-4, Washington State
University, Pullman, WA. and
Northwest On-Farm Test Results. S.B. Wuest, B.C. Miller, R.J.
Veseth, S.O. Guy, D.J. Wysocki and R.S. Karow. 1993. Department
of Crop and Soil Sciences Technical Report 94-1, Washington State
University, Pullman WA.
results from previous year's on-farm tests.
Testing: A Grower's Guide. B. Miller, E. Adams, P. Peterson
and R. Karow. 1992. Washington State University Cooperative Extension
A guide to
designing and carrying out OFT. Includes forms for record keeping.
20 pages, $1.00. Order from WSU Cooperative Extension Bulletin Office