Bayesian MIRT w Log-Normal Prior PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Ryan Walters posted on Thursday, July 12, 2012 - 12:06 pm
I have a question about the use of log-normal priors in Bayesian MIRT recommended by Patz & Junker (1999). When I try to specify this prior, I get the following warning message:

*** FATAL ERROR
THE PRIOR SPECIFICATION FOR PARAMETER 1 IS NOT AVAILABLE.

This warning appears for all priors except normal. I must be missing something, so any insight is appreciated. I've attached my code below.

Thanks!
Ryan

DATA:
FILE IS Wave1.dat;
FORMAT IS Free;

VARIABLE:
NAMES ARE v1 v2 v871 v875 v879 v883 v887 v891 v895 v871r v875r v879r v883r v887r v891r v895r;
USEVARIABLES ARE v871r-v895r;
CATEGORICAL ARE v871r-v895r;
CLUSTER IS v2;
MISSING ARE ALL(-9);

ANALYSIS:
TYPE IS COMPLEX;
ALGORITHM IS GIBBS(RW);
ESTIMATOR IS BAYES;
FBITERATIONS ARE 5000;
THIN IS 100;
PROCESSORS ARE 8;

MODEL:
F1 BY v883r* v887r* v891r* (a1-a3);
F2 BY v871r* v875r* v879r* (b1-b3);
F3 BY v895r* (c1);
[F1@0 F2@0 F3@0];
F1@1; F2@1; F3@1;

MODEL PRIORS:
a1-a3 ~ LN(0,.5);
b1-b3 ~ LN(0,.5);
c1 ~ LN(0,.5);

OUTPUT:
STDYX;
RESIDUAL;
TECH8;

PLOT:
TYPE IS PLOT2;
 Linda K. Muthen posted on Friday, July 13, 2012 - 10:49 am
The LN prior is for variances. See page 620 of the user's guide.
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: