两个IV之间的correlation与它们的interaction effect on DV是两个不同的概念。
我们计算IV之间的correlation主要为了防止Multicollinearity,如果过高(>0.5),是不能够将这两个变量同时运用在同一个model。
计算interaction term是为了验证moderate effect,即一个变量是否影响另一个变量对DV的作用,可以是正,负,或不显著。譬如,X1 and X2 have significant and negative effects on Y, and X1*X2 positively affect Y. It means that X1 has a moderate effect on the relationship between X2 and Y: X1 reduces the negative effect of X2 on Y. You can also say that X2 is as the moderator between X1 and Y. It depends on the way you interpret your model and variables.
就你的情况,IV的main effects不显著,但是各自影响对方与DV的关系。虽然不知道你的研究领域,但这样的结果意义不大。一般而言,人们会在两个变量,或者其中一个变量显著的情况下加入moderator。 |