Whether you raise crops, forages or animals, there seems to be a constant flow of new products into the agricultural marketplace. You see advertisements in trade publications, at stores and farm shows, and on the Internet and television. How do you know if these new products really work? Do they really make your other inputs work better or give you greater yields or better quality in your end product or anything else that they promise?
Marketing of these new products will often include pictures, testimonials and some form of numerical data. Naturally, the numerical data will show an advantage to using the new product. However, it is very important to know some things about the numbers. The most important thing to know is whether or not the numbers are statistically significant. I won't go into an in-depth lesson on statistics, but here are some basics that need to be considered.
Was the study replicated? We all know there is variability in biological systems. Replication allows an experiment to be conducted in a way that accounts for natural variability. This way we can have a certain degree of assurance the difference observed is real and not just due to natural variability.
Was the study conducted one year or over multiple years? Again, there is variability from year to year. We want some degree of assurance that the product will work every year and not just once in a while when conditions are just right.
Was the study done in an environment similar to yours? Even if it was properly replicated over multiple years, results may be completely different in a field or laboratory far from your location. What works in the Panhandle of Oklahoma may not work on the Gulf Coast of Texas and vice versa.
Was the study statistically analyzed? If so, look for a least significant difference (LSD) value. This is, as the name implies, the least numerical difference between any two treatments that is significant. Differences between product A and product B that are less than the LSD are not significant, and we cannot say with confidence that A and B are truly different from each other. Likewise, differences between product A and product B that are greater than the LSD are significant, and we can say with a degree of confidence that A and B are truly different from each other.
In Table 1, varieties A and B are not significantly different since the LSD is 6 and the difference between A and B is only 3. A is significantly different from C and D since the difference between A and C or D is greater than 6. Varieties B, C and D are not significantly different from one another.
Whether you are looking at different wheat varieties, pesticides or any other input, look for replicated data from multiple locations and multiple years in environments similar to yours that show they have been statistically analyzed. Only then can you be confident that the new product will work for you.