Bayesian Monte Carlo
Message/Author
 Danming An posted on Thursday, December 01, 2016 - 3:43 pm
I am running a simulation study for Beyesian SEM. I constrained some priors using model constraint because in this case there's a strong prior belief that some parameters are larger than zero.

I get a few error messages as below in Tech9 output and wonder how to interpret them.

REPLICATION 230:
THE MODEL ESTIMATION DID NOT TERMINATE NORMALLY.
UNABLE TO GENERATE PARAMETERS SATISFYING THE CONSTRAINTS.
CHECK THE STARTING VALUES.

THE PROBLEM OCCURRED IN CHAIN 1.

Thank you!
 Danming An posted on Thursday, December 01, 2016 - 3:46 pm
For reference here's the model setup:

MONTECARLO:
NAMES ARE x1 x2 y1 y2;
NOBSERVATIONS = 100;
NREPS = 1000;
SEED = 1109;

MODEL POPULATION:
[x1*0];
[x2*0];
[y1*0];
[y2*0];
x1*1;
x2*1;
y1*.978;
y2*.978;
y1 on x1*.1 x2*.1;
y2 on x1*.1 x2*.1;
x1 with x2*.1;
y1 with y2*.1;

ANALYSIS:
ESTIMATOR = BAYES;
FBITERATIONS = 1000;

MODEL:
[x1*0](m);
[x2*0](m);
[y1*0](m);
[y2*0](m);
x1*1(varx1);
x2*1(varx2);
y1*.978(residy1);
y2*.978(residy2);
y1 on x1*.1(a1);
y1 on x2*.1(p1);
y2 on x1*.1(p2);
y2 on x2*.1(a2);
x1 with x2*.1 (rx);
y1 with y2*.1 (ry);

MODEL PRIORS:
a1~N(0,.05);
p1~N(0,.05);
a2~N(0,.05);
p2~N(0,.05);
rx~N(0,.01);
ry~N(0,.01);
m~N(0,0.01);
varx1~IG(.001,.001);
varx2~IG(.001,.001);
residy1~IG(.001,.001);
residy2~IG(.001,.001);

model constraint:
a1>0; a2>0;
p1>0; p2>0;
 Bengt O. Muthen posted on Friday, December 02, 2016 - 11:42 am
We request that Discussion postings fit into a single window. When longer postings are needed like here, please send the output to Support along with your license number.
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: