The first critical step in research is to keep it simple.
“Ask one question, and keep it very specific,” said Farming Smarter researcher Lewis Baarda.
“It’s important to keep that as simple as you can and have one research question where you can look at one thing at a time. The simpler you make it, the easier it is to not have variability come in and screw things up.”
Reducing variability is next — you want to know any effect you’re seeing is a result of management, not variations in the field.
“If you have a field with lots of hills, it might not be the optimal field to use for on-farm research,” said Baarda. “You want a field with the least variability, a flat field that you get relatively equal yields across the field all the time.”
That’s where producers can run into trouble, said Kevin Auch, a veteran at conducting on-farm research.
“We think that we’re testing for one treatment and we’re actually throwing a bunch of other noise in there that’s throwing off what we’re doing,” he said.
“If you’re testing for optimal rates of fertilizer and you’re going through variable parts of your field, you’d better have a good way of measuring that with some degree of accuracy before you can actually draw conclusions from what your treatment is.”
A control or check strip can eliminate some of that noise, said Baarda.
“Having that check strip or control that you haven’t done anything to gives you a bit of a bearing of what’s going on,” he said. “And then you know that whatever results you see are a result of the experiments you’ve gone through.”
Replicating the trial in different parts of a field will also help.
“When you don’t have your trial repeated somewhere, it’s tough to say exactly if the effect was because of what you did to it because you only did it once.”
Baarda also encourages producers to use statistics wherever possible.
“It’s pretty easy to give it the old eyeball test, but statistics will add a little bit of rigour to it,” he said. “It can be intimidating, but when it gets down to it, it’s plugging a few numbers in, and it’s not too complex.”