Home » General » Learning from Strip Trials.

Learning from Strip Trials.

This article is written by Dr. George Rehm, University Minnesota, Soil Fertility Specialist (retired).
See more of Dr. Rehm’s blogs at agwaterexchange.com. 

Use of  strip trials as a learning as a way to learn is becoming more popular across the Corn Belt.  This is to be expected.  Crop producers have a thirst for information.  With GPS technology and yield monitors, and the use of common sense, it’s not difficult to establish strip trials for the purpose of evaluating a concept or compare one or more products or rates of a product.  There are, however, some important considerations for the conduct of a strip trial.  These begin with planning before planting and continue with appropriate interpretation of the data following harvest.  These considerations are summarized in the paragraphs that follow.

IN THE PLANNING PROCESS, SIMPLICITY RULES — Speaking from years of experience, when planning, it’s very easy to bite off more than you can chew.  What looks easy or simple on paper can be a logistical problem when you go to the field.  So, make comparisons simple.  If comparing rates of nitrogen fertilizer for corn, for example use no more than three rates.  It’s nice to have a control  (the variable of interest is not used).  The treatments to be compared must be repeated in the field  at, least  three times.  If comparing rates of nitrogen fertilizer for corn, for example, use no more than three rates.  It’s nice to have a control (the variable of interest is not used).  The treatments to be compared must be repeated at least three times.  The replication must be in the same field.  It is almost a waste of time if fields are used as replications.  If a control is used, it should also be replicated three times.

SITE UNIFORMITY — The day of selection of the site for a strip trial is probably the most day for the entire project.  Soil uniformity is a must.  There is no easy and simple procedure that can be used to correct for lack of soil uniformity at the site.  There are several tools that can be used to select for soil uniformity.  The Soil Survey should not be ignored.  Soil test information based on either grid or zone sampling can also be very valuable.  Time spent in selecting a uniform site is time well spent.

PRODUCTION PRACTICES — Once a specific comparison has been selected it’s very important to keep other production practices constant.  For example, information from a strip trial designed to compare nitrogen rates has little value if varieties are changed in the trial area.  Except for the factor of interest, keep all other production practices constant across the strip trial area.  Two production practices that change across the strip trial cannot be changed at the same time.  Careful planning for this type of project takes time and thought.

DATA COLLECTION — Unless there are special reasons to do otherwise, samples collected from treatments at any strip trial site should be collected at the same time.  This practice reduces variability in the data.  Considering yields, use of  combine yield monitors or weigh wagons is certainly appropriate.  Although this may be obvious to most, it is essential to record yields from each strip separately.

STATISTICAL ANALYSIS — There’s a reason for repeating (replicating) each treatment at least three times.  The project is not complete until the data collected have been analyzed with a mathematical procedure called “statistical analysis”.  I think that we all realize that there is variability across any field.  With all factors being equal, we could combine four strips across any field and the yields would not be the same.  So, when we see differences in yield, the obvious question is: “Is the difference in the yield the result of a real difference caused by the factor being considered or variability across the field?”  Statistical analysis is the tool needed to answer this question.  There is no other way to answer this question.

Let’s look at an example illustrating the importance of statistical analysis.  Using strip trials in different counties, two rates of nitrogen were compared.  There were three strips of each rate.  For a field in Kandiyohi County with corn following a soybean crop, yields from the lower nitrogen rate (149 lb. soil + fertilizer N/acre) were 123, 157, and 170 bu./acre for the three strip receiving this rate.  These three yields average to 150 bu./acre.  For the higher nitrogen rate (199 lb. soil + fertilizer nitrogen), the three yields were 157, 176, and 166 bu./acre.  This averages 171 bu./acre.  Using these arithmetic averages, the initial conclusion is that the higher nitrogen rate was better than the lower nitrogen rate  It would certainly appear that 171 bu./acre is better than 150 bu./acre.  If statistical analysis is used, however, the difference in yield is not statistically significant.  Why?  This conclusion is the consequence of substantial variability among three replications.  In other words, the arithmetic difference is due to variability in yield across the field rather than the factor being compared.

For the same project, a strip trial was used on a field in Carver County.  The corn/soybean rotation was used.  The low nitrogen rate was 102 lb./acre and the higher nitrogen rate was 151 lb./acre.  Yields from the three strips with the low nitrogen were 181, 196, and 195 bu./acre with an average of 191 bu./acre.  For the high nitrogen rate, yields from the three strips were 208, 210, and 207 bu./acre with an average of 208 bu./acre.  Statistical analysis of this yield data showed that the difference between 191 bu./acre and 208 bu./acre was not due to variability in the field.  It was, in fact, the result of the rate of nitrogen applied.  Notice that variability among the three replications for each nitrogen rate was small.  Thus, we can say with confidence that there was a REAL difference in yield caused by the rate of applied nitrogen.

Nearly everyone involved with strip trials wants to present an economic analysis of the yield data.  This is logical.  HOWEVWE, an economic interpretation is only valid if differences between or among treatments is STATISTICALLY SIGNIFICANT.  Otherwise, we make a serious MISTAKE that could have serious economic consequences.  For the Kandiyohi County field, the difference in yield could have been caused by treatment applied or natural variation in the field.  We have no way of knowing the real cause.  For the trial in Carver County, we are sure that the difference in yield was due to the rate of nitrogen applied.  Use of statistical analysis allows us to reach this conclusion.  Now economic interpretation can be applied to the results.

SUMMING UP — Use of strip trials is a good  way to make comparisons between or among factors that affect crop production.  In addition, these comparisons can be conducted in growers’ fields.  However, it’s not an easy task to do an accurate job.  Good planning is needed at the beginning and STATISTICAL ANALYSIS is essential at the end.  There are too many comparisons where statistical analysis is ignored and only arithmetic averages are used.  Without statistical analysis, there can be any number of interpretations of the data.  Statistical analysis eliminates the potential for confusion.

Dr. George Rehm,
University of Minnesota
Nutrient Management Specialist (retired)
rehmx001@umn.edu


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: