论文部分内容阅读
在方差分忻中有三个基本假定:1.每一数值由若干部分相加而成;2.其误差呈正态分布;3.各组具有相同的方差.如果要分析的资料其总体与这三个假定条件的一项或几项不符,则使用方差分析就失去了依据.医学上常用的百分率资料,一般都属于二项分布.在二项分布中,бp~2=(P(1-P))/n,方差随着率而变化.如果几组率要作比较,它们的率不同,方差也就不同,因而也不能用方差分析来进行比较.
There are three basic assumptions in the FD: 1. Each value is summed by several parts; 2. The error is normally distributed; 3. Each group has the same variance. If the data to be analyzed is overall with this If one or several of the three assumptions do not match, the use of variance analysis loses its basis. The percentage of medical data commonly used is generally a binomial distribution. In the binomial distribution, бp~2=(P(1) P))/n, the variance varies with the rate. If several sets of rates are to be compared, their rates are different and the variance is also different. Therefore, it cannot be compared by analysis of variance.