论文部分内容阅读
在茶树育种中,常常根据各种性状(原因因素)与产量、品质及扰逆性(结果变量)间的相关性来选拔有希望的个体或类型。这些相关性往往采用简单相关系数或多元回归法分析。简单相关系数只有在各种性状相互独立时才能反映性状对于结果变量的重要程度,而实际中性状间以及性状与结果变量之间都存在复杂的相关关系,形成一个复杂的网络系统。这样,简单相关系数往往容易掩盖事物的真实面目,不能全面考察性状间的相互关系,造成试验结论的片面性。多元回归分析虽可部分地避免这一弊病,但因偏回归系数的某些不足,使原因对结果的效应不能直接进行比较。 1921年,S.Wright提出通径系数(Path coef-
In tea tree breeding, the selection of promising individuals or types is often based on the correlation between the traits (causes) and yield, quality, and disturbance (outcome variables). These correlations are often analyzed using simple correlation coefficients or multiple regression. Simple correlation coefficient reflects the importance of trait to the outcome variable only when the traits are independent of each other. However, there are complicated correlativity between the actual trait and the trait and the result variable, forming a complex network system. In this way, the simple correlation coefficient is often easy to cover up the true face of things, can not fully examine the interrelationship between traits, resulting in unilateral test conclusions. Although multiple regression analysis can partly avoid this disadvantage, some of the deficiencies of the partial regression coefficient make it impossible to directly compare the causes and effects of the results. In 1921, S. Wright proposed Path coef-