I'm looking into testing a new product next year on my field. What do I need to know to get started? — Justin Schultz
On-farm trials are a great way to test new products, equipment, technology and practices. But, to gather useful, accurate information, even a simple on-farm trial requires good planning and proper execution.
"Start with a testable research question," says Dr. Jeff Coulter, extension agronomist at the University of Minnesota. "What is your question, and can it be answered with the type of trial you have in mind?"
The best way to get reliable data is to keep trials simple, by having only one variable (e.g., crops treated with a product versus non-treated).
"It's easy to get excited and take on too many treatments," Coulter says. "It's always important to keep it manageable so that at the end of the year, you've got results."
Then, after determining the objective of the trial, identify what information needs to be gathered prior to the trial (e.g., crop, planting date, fertilizer or pesticide applications, etc.). This information will help when coordinating replications.
"One important factor is to replicate your treatments," Coulter explains. "Sometimes, something happens during the season and you lose data from some plots, so it's a good idea to start with at least four replications of each treatment. By the end of the study, you need a minimum of three to four replications of data in order to perform a good statistical analysis. It is important to replicate all treatments, including the control. A control or check treatment is one in which the product has not been used so there is something to compare plot results to."
Next is field selection. Where will trial plots be located? Ideally, each replication will be uniform in soil type, slope, drainage, fertility, etc. The goal is to reduce inherent or applied variation in the whole plot to ensure observed treatment differences are real, and not due to unrelated underlying variations.
"Make sure to randomize your treatments within a replication," Coulter advises. "This means to put the treatments out in a random order that is different for each replication. Lack of randomization can create a bias."
When doing a rate study, where yield is measured in response to the rate of an input, make sure to cover a wide range of rates ranging from those that you think are suboptimal to supra-optimal. This creates a yield response curve.
"If there is a yield response to the rate, one typically fits a curve to the data and then calculates the economic optimum rate," Coulter says. "When doing a rate study, in addition to having a low-rate or zero-rate control treatment, it's also important to have a very-high-rate treatment as well. This provides a full view of the yield response curve and ensures that yield is maximized within the range of rates used."
During the growing season, it is critical to consistently scout the plots. Make sure to watch for any possible variations or complications. Take note of what you're observing so that it can be factored in when comparing data later, and be prepared to abandon the plot results if it is obvious that there is something wrong with the overall replication site.
"At the end of the year, when looking at the results, make sure to not just look at the averages," Coulter says. "Average is just average. It's important to dig into the data and spend time thinking about the variability in responses. This can help one micromanage fields. Using a site–specific approach to inputs increases the likelihood of dollars being invested more wisely in all parts of a field, not just on average across a field. This can enhance overall farm profitability and environmental stewardship."