On-Farm Testing in Manitoba - Manitoba Agriculture, Food and Rural Initiatives

On-Farm Testing in Manitoba - Manitoba Agriculture, Food and Rural Initiatives

As a sales agronomist I often get into Demo plots to show farmers something new or something new to them. Demo and plots done by farm supplies and companies are not always real solid data. A lot of things are not controlled . All you are really doing is giving people a "look". I mean there is yield data and observations, but to extract valid information that separates the "other" variability from the effects of tested practices you need more

"More" can generally be broken down into 3 components:

1. Replication - The differences caused by your tested practice, will become evident if you make repeated comparisons. In other word replication helps you determine if differences between plots are due to tested practices or natural variation. 4 replications gives the best chance of success.


2. Randomize - Mix 'em up. Where practices are placed must be fair or "unbiased". Many locations consciously or unconsciously give one or the other an advantage. Something as simple as a coin flip makes things random


3. Statistical Analysis - You have to " do the math" to detect if your tested practices give "significant" differences. In other words if the A canola yield only 1 bu. more over B canola , does it mean anything? The formulas for this give me a headache just thinking about them. Analysis of Variance (ANOVA) or Least Significant Difference (LSD) or Probability Level ( no initials for some strange reason)

With today's gps and mapping equipment and yield monitors this kind of test can be done. You do need planning and patience, but trials can be planted and harvested in very controlled conditions with present on farm technology. Plus a great deal of information can be collected while doing the operations. Now all the producer has to do is add some planning and math to the results. Finding the time will be a lot harder.

The post is a MAFRI factsheet I've found helpful

I've got to say when companies do "real" work in plots they do a good solid job. But remember if you are looking at a "demo" or a real test.