缺失数据配伍组设计方差分析的回归分析方法

本文介绍一种缺失数据配伍组设计方差分析的精确方法——回归分析方法,该方法不同以往使用的估项方法,其实质是把方差分析模型视为一般线性模型,采用边界条件法处理,也适用于配伍组设计方差分析非缺项情形。我们将回归方法和估项方法同原来未缺项的方差分析结果对比,认为回归方法比估项方法准确,尤其在处理组结果上,回归方法比估项方法更准确。

Regression Analysis Method of Analysis Variance Randomized Blocks for Missing Value

In this paper,a accurate method of randomized blocks variance analysis for missing value is introduced-regression analysis method of variance analysis. This is different from estimating value method used before. Its essence is to regard the model of variance analysis as common linear model, and adopt side conditions method for treatment. It is also appropriate to randomized block variance analysis for no missing value. We have compared regression method and previous estimate method with result of variance analysis of original complete value.We consider that the regression method is more accurate than estimate value method, especially to the result of treatment.