Can we test a multilevel moderation effect without significant τ of the unconditional model?

[Chu-Ding LING]

Hello list,

 

Is there anyone here who is expert in HLM? I have a basic question:

 

I am attempting to test a multilevel moderation model, in which the independent and dependent variables are at level 1, while the moderator is at level 2. According to what I learned from the textbook previously, I should run the unconditional model firstly to examine whether the variance component (τ) of the dependent variable’s slope is significant. Unfortunately, it was not significant. It seems that it is not necessary for me to test further. However, I am curious about the result so I did the multilevel moderation test and found it was significant.

 

Now here comes my question: Maybe it is inconsistent with the principle of HLM to test such a moderation test, but how to explain the significant moderation effects since I got the statistic results?

 

I would be grateful if any of you can give me some hopeful suggestions. 

 

---

Chu-Ding LING
Ph.D. Student of Business Administration
School of Management
Zhejiang University

[Wangchj04]

A quick conclusion is the main effect is not significant in the null model. Although it becomes sig in the full one, it could be error one, and the impact size maybe trivial. 

Sent from my iPad

--
-- 本小组推荐：
http://www.statmodel.com/
http://quantpsy.org/
http://www.kthau.com/?page_id=81
http://www.caldar.org/html/2010%20Speakers/mcardle.html
http://courses.nus.edu.sg/course/psycwlm/internet/
（4 TEACHING------4.2 Some notes from previous workshops）
http://blog.sina.com.cn/u/2142257021
http://lavaan.ugent.be/
http://personality-project.org/r/r.guide.html#Intro
 
 
You received this message because you are subscribed to the Google
Groups "Mplus学习小组" group.
To post to this group, send email to mpl...@googlegroups.com
To unsubscribe from this group, send email to
mpluser+u...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/mpluser?hl=zh-TW
---
You received this message because you are subscribed to the Google Groups "Mplus学习小组" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mpluser+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[ya]

哈喽阿，

 

你的意思是，

 

1. 你的unconditional model是指的null model还是带level－1 covariate的model？

一般情况下，你需要先做一个null model，看level－2是否有显著影响，说白了，就是先检验是否有必要考虑multilevel structure，通过这个模型，你可以得到ICC。然后，根据你做的领域里大家一般认可的一个值的大小来决定ICC是否足够大，如果够大了，就需要考虑multilevel。

 

2. 在null model的基础上，你才开始纳入level－1的covariate，有了covariate，才会有slope的问题。然后接着是level－2的covariate，也就是你的moderator。你后来得到的显著的结果，说明的是moderator的slope显著，所以调节作用应该是显著的。而且这个跟你null model给出的level－2是否有显著影响应该是不矛盾的。所以，貌似是你的第一个分析出了问题？

 

3. 关于这方面的内容，中文的书可以看张雷的那本多层线性回归。英文的资料可以google Kristopher Preacher.

 

4. 提问的话，最好是贴上你采用的程序（或相应的报错），这样大家比较容易看出存在的问题来：）

 

good luck

 

 

---

ya

 

发件人： Chu-Ding LING

发送时间： 2014-05-18 07:42

收件人： mpluser

主题： [mpluser] Can we test a multilevel moderation effect without significant τ of the unconditional model?

--

[凌楚定]

​抱歉，回复迟了。问题已解决，谢谢您！

​

Chu-Ding Ling