Published at the Chat 'n Chew Cafe, July 2000

Tips for Test Plots

R.L. (Bob) Nielsen
Agronomy Dept., Purdue Univ.
West Lafayette, IN 47907-1150
Email address:

Conducting agronomic field research is often complicated by the ‘slings and arrows’ of Mother Nature. The goal of most on-farm field plot studies is to identify differences among ‘treatments’ under ‘real world’ conditions. These ‘treatments’ may be corn hybrids, fertilizer rates, plus or minus fungicide, seeding rates, tillage practices, etc. The ultimate variable of interest is usually grain yield. The ‘real world’ conditions contribute to additional yield variability and often make it difficult to evaluate the true ‘treatment’ effects.

If the yield variability caused by ‘real world’ factors were distributed equally among the test plots, then ‘treatment’ comparisons could still be made confidently. When the ‘real world’ influences test plots unevenly, then ‘treatment’ comparisons become more confusing. In other words, you run the risk of incorrectly attributing a yield effect to a treatment when in fact it may have been unduly affected by a chance occurrence of an unrelated factor that influenced yield positively or negatively relative to other plots in the trial.

The upshot of this if you are conducting a field trial of any kind is that you should be walking the plots throughout the growing season and taking notes on odd things occurring out in the plots. In your role as a ‘researcher’, you have the responsibility to assess the quality of the test plots and the subsequent yield data that will be measured from them. Sometimes, you need to delete individual plots from the analysis of a study if you determine that they have been disproportionately influenced by some stress factor.

For example, let’s say that you are conducting a simple corn hybrid evaluation trial in which you planted ten different hybrids in 12-row strips the length of the field interspersed with a check hybrid (we can debate in a later article whether check hybrids are worth anything). The field is perhaps a typical Indiana field containing a number of different soil types as well as variable topography and drainage. Frequent, heavy rains (technically known as ‘goose-drownders’) early in the growing season resulted in sizeable areas of drowned out corn and other areas of live, but stunted and yellow corn. These areas of dead or stunted corn are distributed rather randomly throughout the hybrid trial, such that the individual hybrid strips contain variable proportions of the soggy soil problem.

Either through extensive field walking or perhaps via aerial photographs, you really ought to try to determine the extent of the problem in each and every one of the 12-row hybrid strips. If some of the hybrid strips are unduly affected by this stress, then they should be considered for removal from the study. There is nothing shameful about tossing individual plots when your thorough field notes indicate excessive yield influence by non-treatment factors.

The advent of GPS-enabled mapping technologies offers growers/researchers the opportunity for accurately defining the boundaries of problems like wet areas within a test plot. Furthermore, some GIS software programs offer the option of editing the individual yield monitor data points. Technically, one could completely alter the test plot data for nefarious purposes. However, a more benign opportunity would be to delete those data points that lie within the mapped boundaries of the problem area, recalculate the average yield for the remaining data points in each ‘treatment’ strip and, thus, allow you to retain plots in a trial whose data would otherwise be suspect.

Corn Growers GuidebookFor other information about corn, take a look at the Corn Growers Guidebook on the World Wide Web at

It is the policy of the Purdue Agronomy Department that all persons shall have equal opportunity and access to its programs and facilities without regard to race, color, sex, religion, national origin, age, or disability. Purdue University is an Affirmative Action employer. This material may be available in alternative formats.
© 2000, Purdue University
End of document