What is a proportionate next step?
Choosing the smallest step that can answer the question. A trial asks the most of a farm: land set aside, time and cost. As the published wheat case puts it, a promising pattern doesn’t always call for a field trial, and many questions need only one or two steps, such as correcting records and measuring what’s missing.
Last time's quick check: Finland measured selenium at every link, from fertiliser to blood, so each step could be seen moving in turn. In Module 2 we left a wheat farm with a striking pattern: grain zinc tracking fibre across five samples, a correlation of 0.98. The map listed rival explanations and gave an honest "not yet".
The published case's plan: correct the records, then measure soil, roots and grain together at thirty matched spots. Why not jump straight to a trial? Because a trial asks the most of a farm: land set aside, time and cost. As the published case puts it, a promising pattern doesn't always call for a field trial, and many questions need only one or two steps. Choosing the smallest step that can answer the question is called a proportionate next step.
Now imagine the sampling study is done. Here are two ways it could turn out. Both are hypothetical, for teaching. Branch one: bran fraction explains the zinc–fibre link. Samples with more bran have more of both, and at the same bran fraction the link disappears. Then the pattern is about the structure of the grain, not the soil, and changing field practice would be the wrong response. The next questions are about variety and milling.
Branch two: at matched spots, soil biology tracks grain zinc even after allowing for bran fraction and yield. That's a stronger lead, but still not a cause: something else about the better spots could drive both. Now it's worth testing a specific practice directly. So how would you test it? With everything from Module 3.
Choose one specific practice, defined exactly: what, how much and when. Randomise it to strips within each field, so no gradient or hidden factor can line up with it. Block by field, because fields differ. Replicate: several strips in every field, across several fields. Measure the outcome you care about, grain zinc, the same way everywhere, and record yield and bran fraction so you can see the route.
Write down the map, the analysis and what would count as a result before the first strip is treated. And decide in advance what you'll do if the answer is "no effect". Even then, one farm and one season give one answer, for that farm in that season.
Repeating the test across farms and years is how a lead becomes knowledge. Next, to finish: what it takes to do all of this well.
Define one practice exactly: what, how much and when. Randomise it to strips within each field, block by field because fields differ, and replicate across several strips and fields. Measure grain zinc the same way everywhere, record yield and bran fraction, and write down the map, the analysis and what counts as a result before the first strip is treated.
One answer, for that farm in that season, and no further. Repeating the test across farms and years is how a lead becomes knowledge. That’s also why a good plan decides in advance, before the first strip is treated, what it will do if the answer is “no effect”.
Causal evidence is changing how farming is measured and rewarded. Leave your email and we will tell you when new films, courses and findings are published.
Prefer to read online? Vitagri also writes on Substack. See our privacy policy for how we use your email.