Statistical Considerations for Evaluating Micronutrient Tests

Abstract

AbstractFor the purposes of making recommendations, micronutrient soil or tissue tests are basically classification procedures. The empirical test variable usually groups samples into three areas, i.e. a group containing mostly samples with no micronutrient problems, a second group containing mostly samples with the micronutrient problems, and a third group containing samples both with and without micronutrient problems. The third group described above can be thought of as occurring at a “transition zone” where the correct status of a sample is unidentifiable by the micronutrient test. Statistical methods using an interaction Chi‐Square are presented for identifying the boundaries of the “transition zone.”It is proposed that a micronutrient test be evaluated in terms of the proportion of successful recommendations. A successful recommendation results when the proper status of a sample is identified by the test. Under the assumption of random sampling from a stable population, the above evaluation scheme follows a binomial distribution. The computation of variances and confidence intervals is shown and discussed for the proportion of successful recommendations. A statistically valid method for comparing different empirical tests, such as extracting solutions, is shown and contrasted to previously used methods.