Conducting On-Farm Research
Researching new practices at home is one of the best ways to see if they will work for you—in this photo, a Lancaster County farmer compared corn performance in planting green and early cover crop burndown. Photo credit: H. Reed, Penn State Extension
Conducting research on your own farm can increase profitability. Specifically, on-farm research can help you determine if a product or practice is worth the expense or time by revealing if it has a positive effect, negative effect, or does not make a difference. On-farm research can be done to fulfill your curiosity on your farm, and it can also be done in partnership with private industry, Extension, or other agencies.
A research trial is distinguished from a demo in its use of statistics. While demos are useful and can show off a new technology or production practice, research trials help you sort out specific questions and arrive at definite conclusions through replication, randomization, and hypothesis testing.
The Sustainable Agriculture Research and Education program released an in-depth technical bulletin that provides a ten-step outline for developing a successful on-farm research project. These steps are summarized alongside quick tips and examples in Table 1.
| Step | Tip | Example | |
|---|---|---|---|
| 1 | Identify your research question and objective | Choose just one yes-or-no question to answer | "Can a legume cover crop substitute for my standard side-dress nitrogen fertilizer application?" |
| 2 | Develop a research hypothesis | Stems directly from the research question; a statement can be confirmed or denied with actual data at the end of the experiment | "A cover crop of hairy vetch before my corn will provide enough N after the pre-emerge application to achieve my target yield." |
| 3 | Decide what you will measure and what data you will collect | Choose what will test your hypothesis; keep in mind cost, practicality, feasibility, necessary technology, special skills needed | yield, leaf chlorophyll content |
| 4 | Develop an experimental design | Includes arranging treatments in the field to reduce error and bias; based primarily on the number of treatments you are investigating* | Paired comparison replicated 4 times; two treatments include hairy vetch cover crop and no cover crop |
| 5 | Choose the location and map out your field plots | Consider field history, as it may interfere with your results; Plots should be easy to maintain, for example, set up plots to run the length of field and wide enough for one or two passes with the combine; allow enough room for buffer rows between plots; try to choose a uniform area to minimize variation; consider adjacent spaces as they can impact the experimental field | Uniform field with no cover crop history chosen close to home farm; plot length= length of field; plot width = 50 ft. (1 combine pass plus buffer rows); see Figure 1 |
| 6 | Implement the project | Establish plots based on map, and mark clearly with flags or stakes; manage plots the same except for treatments; plan ahead and communicate--share map, management plan and calendar with all involved | See management plan example below, Table 2 |
| 7 | Make observations and keep records throughout the season | Rainfall, temperature, events like pest problems and field operations, etc. | "Hailstorm in July; black cutworm appeared to be more active in no cover crop control plots, but numbers did not reach economic threshold; wet spot at southern end of block 3; weed outbreak along eastern edge of block 1 due to spray skip" |
| 8 | Collect research data | Try to label datasheets and collection materials before heading to the field; make sure to keep all plots and treatments separate; use random sampling procedures; for yield, try to sample center rows, leaving border rows between plots | |
| 9 | Analyze the data | Determined by experimental design; for a paired comparison, a simple t-test is used to compare two treatments* | See sample calculation, Table 3 |
| 10 | Interpret the data and draw conclusions | Analysis may reveal significant positive or negative effects, or no significant effect; this may lead you to change future practices, or you may want to repeat the experiment to be sure; it can be helpful to consult an Extension Educator or Specialist, or other trusted advisor to help interpret results | Average difference between treatments was 14.5 bu/A, while the LSD was 22.58 bu/A. Since our difference is less than the LSD, we conclude there is no significant difference between treatments, or that hairy vetch provided enough nitrogen to meet corn needs equal to our side-dress N supplied to the no cover crop treatment. |
| Date | Task |
|---|---|
| 14-Aug | Flag out plots |
| 15-Aug | Plant hairy vetch, 25 lb/A |
| 30-Apr | Spray burndown on all plots - 22 fl.oz. glyphosate + AMS |
| 10-May | Plant Corn |
| 13-May | Broadcast UAN on all plots - 50 units N |
| 13-Jun | Chlorophyll meter test (SPAD) all plots |
| Mid-June | Side-dress 75 units N no cover crop only |
| July-Sept | Observations as needed |
| Late-Oct | Harvest plots individually |
Table 3. Sample calculations for hypothetical paired comparison experiment - Can hairy vetch supply corn N needs?
Additional calculations
- Add up column 6. Sum of squares = 605
- Subtract 1 from the number of blocks to get the degrees of freedom. 4-1 = 3
- Divide 605 by the degrees of freedom to get the variance. 605/3 = 201.67
- Divide 201.67 by the number of blocks to get the variance of the means. 201.67/4 = 50.42
- Calculate the square root of the variance of the means to get the standard error. √50.42 = 7.10
- Multiply this answer by the selected t-value (find on p.19 in SARE bulletin). In this case, the farmer wants to be 95% confident that her results are significant, so she chooses 3.18.
- 7.10 x 3.18 = 22.58 (this is the Least Significant Difference or LSD)
- Compare the average difference in column 4 with the LSD 14.5 <22.58 ; we conclude no difference between treatments.
It is important to be realistic and practical with your data collection plan. Be aware that lots of time and effort go into coordinating and executing on-farm research in addition to your usual practices. Aim for completing a few simple measurements you know you can commit to and do them well. This will also help you be mindful of hidden costs for things like specialized equipment or tools, laboratory analyses, and labor for time-consuming activities like soil sampling.
As difficult as it is to put into practice, try to manage your expectations going into the research project. "No difference" between treatments is still a valid result, even though it can be disappointing. Additionally, use caution when drawing conclusions from your data, especially about the relationship between multiple affects you measure or observe. For example, if you planted a cover crop and measured better weed control and higher yield in those plots compared to the no cover crop control, you can't conclude that the higher yield was caused by the reduction in weeds, since that was not the hypothesis tested in the experiment.
Lastly, share your results at a field day! Whatever question you asked and answered with your project, it is highly likely that other farmers in your community have the same question. Sharing your results may inspire others to change their practices and improve their profitability or sustainability; you may even motivate them to try some research on their own farms.
You can apply for a grant to help fund your own research through the Sustainable Agriculture Research and Education program (SARE). Whether you are interested in conducting an independent research project, or if you would like to participate in Penn State Extension's research network, don't hesitate to reach out to your local educator for support and additional resources.
You may also be interested in ...
Grain Drill Calibration
Grazing Corn Stalks with Beef Cattle
Managing Drought Stressed Corn
