並列摘要
When the traditional statistical software is analyzing a linear models, testing whether the factor is significant, all divided by mean squares error, but nested and split plot design are not. So while testing whether the factor is significant or not, we must judge the design and declare test statement, in order to make the correct test. But now there is a new package, as long as determine the design and the judgment of the effect of the factor is fixed or random and there is no need to declare test statement, we can directly make the correct test. In this paper, four different programs are used to analyze the data of the nested and split plot design, and two different estimation methods for the maximum likelihood method and the restricted maximum likelihood method are used to estimate the parameters. Finally, we use the data of the two designs, the student’s grades and the rice yield, to see the differences between the traditional package and the new package.