看过温忠麟老师的《调节效应和中介效应比较与应用》，上面说到如果自变量是连续的，调节变量是类别变量，那么只要将类别变量分类，分别y对x回归，系数差异显著的话，则调节效应显著。

我在实践的时候遇到问题了，系数差异大什么程度算是显著？如果系数差异显著，而系数本身却不显著，那么能不能判定调节效应显著？以下是我做出的结果，大家帮讨论下，看能不能判断调节变量m的显著性。另附《调节效应和中介效应比较与应用》，方便大家讨论

左边是调节变量m=0时ABCDE五个变量的系数和t值，右边是调节变量m=1时，五个变量的系数和t值，请问如何判断这个m变量的调节效应是否显著？如何才能判断两个回归的系数差异是否显著？

M=0                                                                                          M=1

A                                                                                               A

-0.209                                                                                            -0.339

  （-2.052）                                                                                        （-1.820）

B                                                                                              B

2.41E-013                                                                                            6E-012

       （1.2）                                                                                       （-0.775）

C                                                                                              C

0.076                                                                                                     0.065

   （2.401）                                                                                        （0.974）

D                                                                                               D

-1.118                                                                                                  -0.243

   （-1.509）                                                                                          （-0.316）

E                                                                                                  E

-1.764                                                                                                      0.100

     （1.312）                                                                                        （0.076）

   

[此贴子已经被作者于2008-9-6 20:33:40编辑过]