I understand regression and inferential statistics. I am not a statistical genius by any stretch, but I sure do like playing with statistics. The two semesters I had of them in grad school just a few years ago were two of the hardest and most fun semesters of graduate school. I really enjoyed the classes. I enjoyed statistics.
Enjoyed maybe too much. It ruined me on buying lottery tickets and raffle tickets at fundraisers. Yea, I still buy a raffle ticket, but not a hand full, just one. It also gets me in trouble at plot tours, field days and sales meetings with seed and chemical salesmen. I question their sample size, plot design and resulting conclusions.
I also question their yield vs. economic responses. Farmers are hung up on yield. Well that’s alright because everything we do is based on a yield. You get paid on what you grow. But you had better get paid more for that yield than what it cost you to grow it.
And that is where I get in trouble. Below are a couple of examples of selling yield or lack there of over the return in dollars to get that yield.
Example 1 is a yield significant/economic non significant scenario that is pretty common in Ag right now. The treatment to a growing crop results in a 5 bu/ac yield increase. That yield increase is significant, as in the likely hood that the result is caused by the treatment is greater than random chance. So by doing X you get a 5 bu yield increase. Lest say that yield is corn: Corn is $5 a bushel. So you grossed a 5 bu increase at $5/bu for $25/ac. But it cost you $26 to get the 5 bushel increase. So I lost a dollar (-$1) in the end……….I lost money by increasing yield. Then why do it?
Example 2 is a yield non significant/economic significant scenario that is also pretty common today. The treatment to a growing crop results in a 5 bu/ac yield increase. That yield increase is NOT significant, as in the likely hood that the result is caused by the treatment is not greater than random chance. So by doing X you get a 5 bu yield increase that could or could not be caused by the treatment. But more often than not you get a 5 bu yield increase. Lest say that yield is corn: Corn is $5 a bushel. So you grossed a 5 bu increase at $5/bu for $25/ac. But it cost you $10 to get the 5 bushel increase. So I gained or netted $15 in the end……….I gained money by increasing yield an insignificant amount. Then why not do it?
Both cases point to a flaw in our research models in Ag. We assume that the underlying result of any practice we use has to be a significant yield increase. Yes, I would hope anything I do results in a significant yield increase. BUT ONLY IF IT MAKES ME MONEY. I also would like to know if the economic return is significant (or at least greater than 50/50) for any treatment even when the yield may not be significant by a statistical model.
In other words, farming is a business. What is my return on investment for any practice despite the results of a significant yield response? In the end it’s about net dollars earned, not winning the yield trophy.