About APIM problem PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Stella Zac posted on Thursday, March 28, 2013 - 7:54 pm
Hi, I am a new user of Mplus and APIM.

My APIM model is:
MODEL:
qolph2 on closa pclosa;
pqolph2 on closa pclosa;
pqolph2 with qolph2;
closa with pclosa;

Now my results of APIM model is very strange,please notice the coefficients in the qolph2 on closa and pclosa, and the coefficients in the pqolph2 on closa and pcloas. They are only reverse, not difference on the numbers.


MODEL RESULTS

Two-Tailed
Estimate S.E. Est./S.E. P-Value

QOLPH2 ON
CLOSA -0.148 0.250 -0.592 0.554
PCLOSA 0.282 0.239 1.181 0.237

PQOLPH2 ON
CLOSA 0.282 0.237 1.191 0.234
PCLOSA -0.148 0.252 -0.588 0.557

I think maybe it is the datasets' problem or not?

If it is my datasets' problem,my current dataset is as below:

dyad hiv qolph2 closa pqolph2 pclosa
1 1 18 5 19 4
1 -1 19 4 18 5
....
Is there any problem of my dataset?

Looking forward to your reply soon!
 Bengt O. Muthen posted on Thursday, March 28, 2013 - 9:13 pm
It looks like there is a problem in the data in that QOLPH2 and PQOLPH2 are strongly related, perhaps just an exact transformation of each other.
 Stella Zac posted on Friday, March 29, 2013 - 1:08 pm
To my understand to APIM pairwise dataset, qolph2 and pqolph2, closa and pclosa are exactly the same.You can have a look of my dataset:

dyad HIV qolph2 closa pqolph2 pclosa
1 1 18 5 19 4
1 -1 19 4 18 5
2 1 15 5 17 3
2 -1 17 3 15 5
......
If list all data record, all the mean, sd for the qolph2 and pqolph2, closa and pclosa should be the same.

But that's what I learnt from the dyadic data analysis dataset format.

Is there anything wrong with my dataset format for APIM in mplus?

Thanks.
 Bengt O. Muthen posted on Friday, March 29, 2013 - 1:27 pm
An analysis should not have two variables with exactly the same values. Perhaps you are misunderstanding the data arrangement for APIM and should go back to sources describing it.
 Stella Zac posted on Sunday, March 31, 2013 - 9:11 pm
I figured out the previous dataset problem.Thanks for your help. Now I have another question about coefficient comparison in Mplus.

Here is my APIM model, I want compare the effects of actor1 and actor2.

ANALYSIS: type=gen;
estimator=ml;

MODEL:
qoleh2 on emotion(a1)
pemotion gender mar age;
pqoleh2 on emotion pemotion(a2)
pgender pmar page;
pqoleh2 with qoleh2;
emotion with pemotion;

model test:
a1=a2;

I found the coefficient of independent variable is different when I use code "model test".

For example, when I only use the

MODEL:
qoleh2 on emotion pemotion gender mar age;
pqoleh2 on emotion pemotion pgender pmar page;
pqoleh2 with qoleh2;
emotion with pemotion;

The coefficient of pemotion to pqoleh2 is 0.20.

but when I used the previous "model test" code, The coefficient of pemotion to pqoleh2 changed to 0.009.

To my understanding, the results of "model test" should have the similar results with the original model. If not, what's leading these difference?

Thanks and looking forward to your reply soon.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: