On-farm research is more successful if the farmer has a question he would like to answer. Simplify research objective by making the question into a comparison.

2. Design the experiment

Three items to remember when designing the experiment:

(1) Do not have too many treatments, 3 or 4 is about the most, especially for the first experiment;

(2) Need to randomize the treatments;

(3) Treatments must be replicated so the data can be analyzed statistically.

3. Select the field and lay out the plot

Select a homogeneous field where the experiment will be conducted; same soil type, soil salinity, drainage, etc. This will help reduce errors that affect the yield that are not related to the treatments. One method to lay out treatments and replications is shown, left. Treatments can be randomized in blocks. Treatments should not be in the same order in the blocks.

4. Collect the data

Monitor plots frequently during the growing season. Record and date observations in a notebook. Information to be collected: emergence, stand, weed and insect damage, soil conditions, and weather conditions. Crop inputs (fertilizer, herbicide) must be recorded.

Problems may arise that make some strips or locations in strips unusable, such as weed patches or misapplied crop inputs. In many cases, yield from on-farm research is collected with a yield monitor and DGPS. Make sure the combine is calibrated and the DGPS unit is working.

5. Analyze the data

Statistics will determine if the data collected is due to treatment differences or due to chance. Least Significant Difference (L.S.D. ) will show the minimum difference needed between treatment to be considered a real difference and not due to chance.

6. Determine economic feasibility

Determine how much the ‘new’ treatment costs as compared to conventional practice.