Cover Image
close this bookTraditional Field Crops (Peace Corps, 1981, 283 p.)
close this folderAppendices
View the documentAppendix A - Measurements and conversions
View the documentAppendix B - How to conduct a result test
View the documentAppendix C - How to conduct a result demonstration
View the documentAppendix D - How to conduct an elementary statistical analysis
View the documentAppendix E - How to convert small plot yields
View the documentAppendix F - How to take soil samples
Open this folder and view contentsAppendix G - Hunger signs in the reference crops
Open this folder and view contentsAppendix H - Miscellaneous pulses
View the documentAppendix I - Troubleshooting common crop problems
View the documentAppendix J - Guidelines for using pesticides
Open this folder and view contentsAppendix K - Guidelines for applying herbicides with sprayers
Open this folder and view contentsAppendix L - Important planting skill for extension workers

Appendix B - How to conduct a result test

When is Result Testing Needed?

· To test responses to an improved practice under actual farming conditions: Research station conditions are often more ideal or at least different from actual on-farm conditions. What works well under the more controlled situation of the station may be less than satisfactory in farmers' fields where soil and management are likely to be much less than optimal.

· To test responses in different geographic regions

· To measure the profitability of a new practice

· To measure the variability of results: Farmers are just as interested in the variability of benefits from a new practice as they are in the average benefit. A practice that produces large benefits on some farms but little or none on others is unlikely to gain wide acceptance.

The Procedure

· Clearly describe the practice to be tested

· Divide the test region into zones: The work area may have significant variations in soils, rainfall, elevation, farming systems, etc. It is important to divide the testing region into separate zones if they differ enough from each other to warrant separate recommendations. The number of zones will depend on your area's diversity, the complexity of the practice you are testing, and time and budget limitations. In most cases you will be dealing with no more than two to three test zones within a municipality.

· Decide on the number of farms to be included per test zone: Naturally, the more tests and farms that are included per test zone, the more representative the results will be and the more specific will be the recommendation that follows. However, costs will be higher and so will time requirements.

Two factors determine the number of farms that should be included in a test area:

· If high average benefit is expected from the new practice as opposed to the traditional one, fewer farms need to be included than if the average benefit is lower.

· If a large variation is expected in farm to farm results, more farms need to be included than if a smaller variation is expected,


NUMBER OF FARMS TO INCLUDE IN A RESULT TEST

If you expect an average increase over normal yields of:

And if you expect yield variation between farms within the region to be:

Then you should include this number of farms in your test: (10 maximum)

100 percent

Quite variable

6


Fairly consistent

4

50 percent

Quite variable

9


Fairly consistent

5

25 percent

Quite variable

10


Fairly consistent

6

Extension workers ideally should consult an experienced researcher or extension officer in deciding how many farms to include in a result test. If professional advice is not available it may be better to proceed with result tests using less precise sampling methods. The table below is based on a 5001000 farm work area.

· Decide on how long to run a result test: If the expected benefits of the new practice are likely to be significantly related to weather conditions during the growing period, the test should be repeated over several years. This is often the case with tests involving fertilizer use and changes in plant density and tends to be true with most other practices, at least to some extent. Repeat testing is especially indicated if the first trial takes place during an unusual weather year. Long-term weather records can help determine this, but if not available, local extensionists and farmers can be of help.

· Select individual farms: It is important that selected farms be representative rather than "typical". The participating farms should reflect a cross-section of those in the test area so that trial results can be converted into recommendations generally suitable for the entire area. Remember also that you should be just as interested in determining the variation of response among farms as in the average overall response. Farmers do not harvest averages'

Ideally, the farms should be chosed at random, but this is never completely practical due to the limitations imposed by accessibility and farmer cooperation. However, the less the choice of farms is confined to a particular class of farms and the more you choose farms on an "as they come" basis, the closer you will be to achieving a valid representation.

This principle is much easier to violate than one might expect. For instance, it is easier to work with farms close to a road, with familiar farmers or with farms where good results can be expected. Such biases can totally discredit the results.

· Decide what kind of control plot is needed: If the result test is to compare an old practice with a new one, a control or check plot will be necessary. However, if a totally new crop is being introduced rather than a new practice or new variety, no control plot is needed.

· Choose the location and size of the plots: Plot location will depend a lot on the feelings of the cooperating farmer. This is no problem, as long as he or she does not purposely select the best piece of ground on the farm. Random choice is the best method here unless parts of the farm have been subjected to very unusual management practices such as ultrahigh fertilizer applications. Both the test plot and the control plot should be in the same field and preferably adjoining each other. This helps ensure that both plots are subjected to the same variables. In fact, it may be best to avoid using farms where the two plots cannot be located in the same field.

The plots should be large enough so that the usual farming methods can be followed, yet small enough so that the results are clearly visible. The test and the control plot do not have to be the same size. The test plot can be a portion of it serving as the control plot.

· Conducting and supervising the test: The farmer and his or her usual extra workers should handle all the land preparation, planting, weeding, and other operations normally associated with the crop. They-should also apply the new practice themselves under the guidance of the extension worker(s). This assures that the result test is fully representative of actual farming conditions.

Make sure that all variables other than the practice or input being tested are held constant. One common error of both farmers and extensionists is to take better care of the test plot than the control plot. Such preferential treatment can completely invalidate the results.

Documentation is vital. All inputs used should be measured and recorded to the extent possible. Weather data such as rainfall, hail, and unusual temperature extremes should also be recorded if possible along with any visual differences between the test and control plot during growth.

· Collecting Data: No conclusions can be drawn from the result tests until yields have been measured. The goal is to weigh the harvest from the test plot and an equal area of control plot. The extensionists and the farmer should decide on a harvest date and arrangements should be made to obtain an accurate scale. Gross yields from both plots can be measured at the same time and then converted to a kg/ha, 1bs./ acre or other locally used yield standard.

However, you should always obtain a Yield sample prior to the actual harvest date just in case the plots are inadvertently harvested without measuring yields before the agreed upon date. A properly collected random yield sample is usually accurate within 5 percent of the actual yield and is a cheap insurance policy.

· Analysis of the results: Good records are essential to any valid analysis of the results. By far the best way of interpeting the results is to run a standard-statistical analysis of the yield data. You do not need formal training in statistics to do this. Appendix F gives easy to follow instructions for carrying out a statistical analysis which will enable you to determine the standard deviation (a measure of the variability of responses from the average).

Calculate the standard deviation, since it serves as basis for giving realistic yield expectations when making recommendations to farmers.

Reducing the Risk To Participating Fanners

· Subsidizing inputs:

Result tests: There are two schools of thought here. Some extension specialists feel that all new inputs for the test plots should be provided free to the farmer. They feel this makes it easier to find collaborators and also helps assure control over the plots. Others feel that no compensation should be given unless a completely new or unknown input is involved, Much of the choice depends on the economic condition and receptiveness of the local farmers.

Result demonstrations: Inputs should ordinarily not be subsidized unless there is still some uncertainty about the profitability of the new practice, in which case it probably should not be at the demonstration phase anyway.

NOTE: If subsidies are provided, be sure to include the true costs of the inputs when doing a cost/benefit study.

· Reducing the number of farm tests :

Result tests: Reducing the number of tests may make the test results unrepresentative for the area.

Result demos: Reducing their number will not affect the demonstration principle, but may slow the rate of mass acceptance by farmers.

· Reducing plot size:

Result tests: Plot size should be large enough to allow normal growing practices to be followed. Rather than cut plot size below this limit, subsidies should be offered.

Result demos: Let the farmers choose their own plot sizes as long as normal growing practices can be followed.

· Guaranteeing the price or the yield: The extension agency may guarantee a certain yield or market price to a cooperating farmer, perhaps in the form of a purchase contract. This should only be done with result tests. Demonstrations should stand on their own.