Set Up Field Trials Carefully

How to learn without misleading yourself.

Published on: Apr 13, 2009

There's nobody way to find out how a hybrid, variety, or production practice will work on your farm than to test it on your farm. However, unless you test it correctly, you could wind up with a misleading answer at best, and a totally erroneous conclusion at worst.

 

Bob Nielsen, Purdue University Extension corn specialist, has worked with farmers for years, helping them learn how to set up field trials on the farm. Most of these are strip trials, meaning they plant strips clear across the field, then switch to another treatment and run strips of it across the field. Where you plant each strip, how you decide which treatments go where and how well you keep records on what you do can make all the difference in determining whether you find up with meaningul information after harvest next fall.

 

Dave Nanda, president of Bird Hybrids, Tiffin, Ohio, helped coordinate the Corn Illustrated plots for Farm Progress Companies for the past tow years. There are no organized plots this year. However, Nanda helped demonstrate a variety of ways in which data can either be preserved or tainted.

 

Based on experiences from both these researchers and others, here are tips to think about before you decide to do an on-farm plot.

 

  •                     Commit to the time. If you don't want to spend an extra 4 hours to a day to learn this type of information, then don't even attempt it, Nanda says. Yes, ti takes time to empty seed boxes or to lay out exactly where the next strip will go, or when it's time to switch treatments. A trial testing hybrids, varieties or practices is no place for a tractor jockey whose only goal is to get done as soon as possible,.
  •                     Haste makes waste- What happens when you hurry is disaster, Nanda notes. Garbage in-garbage out- the test for the rest of the season will only be as good as it is laid out in advance, and planted according to a plot plan.
  •                     Plan in advance- Plans for the plot should come together while the snow is flying in January, not while the tractor on the planter is running and the sun is shining in later April or early May. By thinking through the plot plan, including which treatments will go where, you can identify possible pitfalls that might influence your results before you even begin the test.
  •                     Replicate if possible- Replication is what begins to tell you if the results you see could be repeated, or if the differences you see are just due to chance and experimental error, Nanda notes. In a strip trial, also called a field-size trial, repeat the test at least once. In small plots, repeating the experiment up to four times is preferred. That allows you to collect enough data to run statistical analysis on the plot. If you can analyze the data in that way, you can feel more confident about which plots were actually affected by the treatment, and which results you could produce again.
  •                     Assigning positions at random- Jeff Phillips, Tippecanoe County Extension ag educator and an avid researcher, uses a computer program to tell him what to plant where in each replication in small-plot trials. That way, he says, he removes as much chance for error as possible. Scientists call it randomizing.
  •                    Make sure the plot is practical- Nanda once worked with a demonstration plot where one/one thousandth of an acre in a pass was in narrow rows. When it came time to apply liquid nitrogen sidedress, he soon discovered there was no way to cross that plot. Urea had to be applied there, introducing another variable. What's more, the farmer had to back over corn in an adjoining plot to go around the solid block. Both things compromised the data.
  •                     Stay on one soil type- This si easier said than done. Going across two soil types or level and then hilly ground introduces more error, Nanda notes. It's always best if you can confine the test to similar soils. Just in case that's not possible, that's another reason why treatments should be assigned at random.
  •                    Do the same things everywhere- If you decide to cultivate one plot for weeds, then cultivate the entire plot- all treatments- whether they're weedy or not. Doing something to part of the plot and not the whole plot introduces a variable that might affect results, even if it seems logical or not.
  •                     Mark locations carefully- You may think you can remember exactly how you set up the plot, but time has a way of playing tricks on you. Mark the plots with flags. Know which way the treatment goes from the flag- to the left, right, or in the center. Record the plot in a notebook, laid out to scale, with each treatment and the number of rows per treatment marked.
  •                     Leave border rows between treatments- Especially if you're testing nitrogen fertilizer or fungicides, border rows are essential. Otherwise, the treatment may bleed over into the next plot, affecting the results. Typically, border rows are not weighed at harvest as part of the test, Nanda notes.