Save estimates of a DSEM analysis wit...
Message/Author
 Julian Rubel posted on Tuesday, July 17, 2018 - 3:49 pm
Hi,

I was trying to save the estimates of a DSEM analysis for subsequent monte carlo based power analysis. Unfortunateley, I get the following error message:

*** WARNING in SAVEDATA command
Saving of ending values for the ESTIMATES option is not available for
models with covariates. Request for ESTIMATES is ignored.

Is there a solution or some quick workaround for that issue? I would leave the covariate out, but this is acutally the effect of interest.

Many thanks in advance!
Best,
Julian
 Tihomir Asparouhov posted on Wednesday, July 18, 2018 - 10:09 am
I would recommend
output:svalues;
I gives you the model you need to put in model montecarlo. The covariate sample statistics should be computed separately.
 Julian Rubel posted on Wednesday, July 18, 2018 - 3:54 pm
Many thanks!
Ill try that!
 Julian Rubel posted on Thursday, July 19, 2018 - 4:40 pm
This worked! Thanks!

I do have a follow up question:

What if my covariate is a binary variable with 60/40 distribution?

How can I generate this binary variable making sure the distribution is 60/40 and it follows the population model specifications?

Best,
Julian
 Julian Rubel posted on Thursday, July 19, 2018 - 4:42 pm
Sorry, I forgot to say that it is a between-level covariate.

Best,
Julian
 Tihomir Asparouhov posted on Friday, July 20, 2018 - 9:37 am
Look up the command CUTPOINTS in the User's Guide. Here is an example

MONTECARLO: NAMES ARE y w;
NOBS = 10000;
NREP = 10;
NCSIZES = 1;
CSIZES = 100(100);
lagged=y(1);
between=w;
cutpoints=w(0.25);

ANALYSIS: estimator=bayes; TYPE IS TWOLEVEL;

MODEL MONTECARLO:
%within%
y*1; y on y&1*0.4;
%between%
y*0.9;[y*2];
w*1; y on w*1;

MODEL:
%within%
y*1; y on y&1*0.4;
%between%
y*0.9;[y*2]; y on w*1;
 Julian Rubel posted on Tuesday, July 24, 2018 - 12:26 pm
Thanks for your help!
Everything worked as wished!
Best,
Julian
 Julian Rubel posted on Tuesday, July 24, 2018 - 12:41 pm
In the output of the monte carlo analysis of a DSEM model with estimator=Bayes:

Is the "%sig. coef." calculated from a frequentist or Bayesian perspective (the latter meaning that this is equal to the percentage of times the CI interval did not include 0?).
If it is frequentist only, is there a possibility to get also the results of the Bayesian way?

Many thanks!

Best,
Julian
 Bengt O. Muthen posted on Tuesday, July 24, 2018 - 2:47 pm
The results in that column are based on the Bayesian CIs.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a public posting area. Enter your username and password if you have an account. Otherwise, enter your full name as your username and leave the password blank. Your e-mail address is optional.
Password:
E-mail:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: