请教这种情况可否用路径分析，多谢

wmcheng2006 发表于 2013-1-31 12:26
不一样的，一个是中介一个是调节。binary variable 用Mplus可以方便的处理

多谢您的回答。Mplus貌似版上就有这个软件呢
那么这个交互作用能不能在路径分析中表示出来呢？
原来用混合模型检测出来有这个交互作用（也是文章的hypothesis中就提出的），因此这个交互作用对文章来说很重要
不甚感谢！！！

可以

wmcheng2006 发表于 2013-2-1 17:36
可以

这个交互作用就另设一个变量，然后和其它变量一样的扔在路径分析图中就可以了？？
抱歉打扰你！非常感谢！！

baixiangnju 发表于 2013-1-31 05:33
非常不好意思，可以再问个问题不？
路径分析中的间接影响和多元回归中的相互作用A*B是一个意思不？
非常感 ...

There are different. For example, A->B->C. In it, A has an indirect effect on C.
A*B means the interaction effect.

woodhaven 发表于 2013-2-6 04:39
There are different. For example, A->B->C. In it, A has an indirect effect on C.
A*B means the in ...

Thanks for your reply!

Could you be so nice to answer my another stupid question, Please?
For recursive path analysis, the coefficients can be calculated using regression models for every endogenous variable? And we can just put all regression coefficients on path diagram?
Your help will be greatly appreciated!

baixiangnju 发表于 2013-2-11 17:45
Thanks for your reply!

Could you be so nice to answer my another stupid question, Please?

Yes, since the path analysis is a direct extension of multiple regression. You can run separate regression models to get the path coefficients.

A better way is to run just one structural equation model using AMOS or MPlus. In this way, you don't need to run separate regression models.

woodhaven 发表于 2013-2-17 09:19
Yes, since the path analysis is a direct extension of multiple regression. You can run separate re ...

Thanks for your patient answers.
As you said, Amos currently can only handle continuous variables. Thus, I think I still need to calculate path coefficients seperately.
Thanks again!

baixiangnju 发表于 2013-2-18 02:44
Thanks for your patient answers.
As you said, Amos currently can only handle continuous variables ...

You are welcome. My above suggestion is just for the continuous endogenous/dependent variable. If the outcome variable is binary, it will be a different story. Please see the link below. http://www.nrhpsych.com/mediation/logmed.html  . It provides the SPSS syntax at the end.

An easier way is to use Mplus.

woodhaven 发表于 2013-2-18 08:57
You are welcome. My above suggestion is just for the continuous endogenous/dependent variable. If  ...

Thanks for your help!
You mean that the Mplus is able to handle binary response virable?

I once read a paper which suggests to use fully standardized coefficients in path analysis with binary response.  But it doesnot tell which software can solve this.
Thanks a lot!!

不好意思，请问楼主文章发表了没有？我现在写的一篇文章要分析的问题和您的问题很类似（不过不是指某疾病是否发生），能否给个链接参考一下？