“下限不在内”也应是一个统计原则

统计分组理论是统计学的重要组成部分。本文对现行的连续型变量确定组限的“上限不在内原则”提出了质疑。在统计分组中,确定连续变量的组限时,相邻组的组限必须重叠。为了防止这种重叠不致引起混乱,要根据所研究现象的不同特点进行归组整理:对于正指标即越高越好的现象,按“上限不在内”原则;对于逆指标即越低越好的现象,按“下限不在内”原则。

"Lower Limit Is not Included" Should Be a Principle of Statistical Data Grouping

The theory of statistical classification is one of the most important parts of statistics. In this paper,it is doubted that the principle currently in effect, "upper limit is not included", which is to determine class limits through continuous variables. In statistical classification, when determining class limits of continuous variables, class limits of neighbor classes should be overlapping. In order to prevent the confusion caused by overlap, we have to divide the continuous variable into classes and put it into order according to the phenomenon's characteristics, the principle of "upper limit is not included" should be used on positive index, the appearance which is the higher the better. The principle of "lower limit is not included" should be used on negative index, the appearance which is the lower the better.