A two treatment trial where two treatments are paired together in adjoining strips is a type of a randomized complete block design. This trial had ten paired strips, each containing one strip each of the two treatments.
Corn Harvest Analysis
This type of trial can be analyzed using a t-test, which determines if numerical difference between treatments are legitimate. Following is a "cookbook" approach to analyzing the data from a trial with up to ten double strips, each containing one strip each of the two treatments. The second and third columns are labeled "yield," but any type of data can be analyzed this way.
The difference between the two treatments is given in Cell 3. The number in Cell 8 is called the "t value" (be sure to record it without the negative sign, if Cell 3 us a negative number, that is, if treatment B has a higher yield than treatment A). We assess whether the difference between treatments (Cell 3) is real or whether it's due to random chance using the following test:
We're 90% sure the treatments really differ if Cell 8 is larger than: 1.860
We're 95% sure the treatments really differ if Cell 8 is larger than: 2.306
Therefore, the statistical confidence that there is a difference between these treatments is greater than 95%. So, the plots treated with SO4 produced a significantly higher corn yield than those that were not treated.
Soybean Yield Analysis
For the soybean analysis, the statistical confidence appears to be less and the resulting conclusion is that the difference is not explained by the statistical model, meaning that there is not a real difference between the two yield values, despite the roughly 3 bu/A increase when using gypsum.