运用正交表各列矩阵像的分解技术,将各列平方和分解成相互独立的若干部分之和.利用这些分解得到的部分平方和,不仅可以解决饱和模型的统计分析问题,还能够对列显著但某些子成分不显著的列增加新的约束条件.在考察列效应的一般约束条件的检验问题时,文章通过对一般约束条件进行转换,使其变成一个成分分解问题,再利用成分分解的相关结论,完满地解决了这一问题.
The sum of square in each column in each column was divided into the sum of some parts independently by using orthogonal decomposition of matrix images. Not only we can do statistics analysis about saturated model, but also we can get some new constraint equation if the column is significant but some decomposition is not significant. For the testing of general constraint equation of some column,we perfectly solve this problem by transforming the constraint equation to a decomposition of column effect.