An Evaluation of Soil Test Information in Agricultural Decision Making
利用田间地块数据估计土壤硝酸盐浓度的后验分布，优化施肥决策，发现土壤测试可使施肥量减少15%至41%，但忽略不确定性会高估测试效益。
The value of soil-test information in planning fertilizer application levels is determined by using agricultural field-plot data to estimate the posterior distribution of mean soil-nitrate concentrations at a give location. Optimal decisions concerning fertilizer application levels are made with respect to this posterior distribution. Average reductions in fertilizer application rates range from 15 to 41 percent, depending on the form of prior information that is available. These reductions are achieved by increasing the variability of application rates over time. Disregarding the uncertainty that remains after the soil testing significantly overstates the expected benefits of soil testing.