关于两条回归直线截距或高度差异的统计学意义检验

Some Comments on the Methods of Significance Test between Intercepts or Elevations of Two Linear Regression Lines

两条回归直线截距间差异的统计学意义检验与两条回归直线高度差异的统计学意义检验常含混地被认为是一回事。在统计书籍上有四种检验方法,对同一实例计算结果竟完全不同。研究结果得出以下结论:一种方法是对两条平行的回归直线纵向高度之差的检验,另一种方法是检验两条回归直线截距间的差异,而与此两条回归直线是否平行无关。另两种方法找不到任何统计学上的依据。

The significance test between intercepts of two linear regression lines used to be considered ambiguously as the same thing as the significance test between the elevations of these two lines. The same example of two linear regression lines was analysed for such test by four methods which bad been found in the statistical books. The results were totally different. The reason was studied and it was concluded that these two tests had different meanings. The first method is a test for the difference between elevations of two paraller linear regression lines and the second one is a test for the difference between the intercepts of two linear regression lines.