The article discusses a method to make decisions about variance components in a statistical model. The researchers developed a way to estimate these components using a technique called kernel estimation. They found that their method is very accurate, with error rates decreasing as more data is collected. An example is provided to illustrate the effectiveness of the approach.