Latent 3-way interaction in Monte Carlo
Message/Author
 Kelsey Borner posted on Monday, May 11, 2015 - 3:35 pm
"THE POPULATION COVARIANCE MATRIX THAT YOU GAVE AS INPUT IS NOT POSITIVE DEFINITE AS IT SHOULD BE."

MONTECARLO:
NAMES ARE P1 - P3 C1 - C5 W1 - W3 PA1;
NOBSERVATIONS = 150;
NREPS = 1000;
ANALYSIS:
TYPE = RANDOM;
ALGORITHM=INTEGRATION;
MODEL POPULATION:
PRAC1 BY P1 - P3*1;
CON1 BY C1 - C5*1;
WAR1 BY W1 - W3*1;
PRACWAR | PRAC1 XWITH WAR1*.5;
PRACCON | PRAC1 XWITH CON1*.5;
CONWAR | CON1 XWITH WAR1*.5;
PRACCONWAR | CON1 XWITH PRACWAR*.5;
PRAC1@1;
CON1@1;
WAR1@1;
PA1 ON PRAC1*.5;
PA1 ON CON1*.5;
PA1 ON WAR1*.5;
PA1 ON PRACWAR*.5;
PA1 ON CONWAR*.5;
PA1 ON PRACCON*.5;
PA1 ON PRACCONWAR*.5;
MODEL:
PRAC1 BY P1 - P3*1;
CON1 BY C1 - C5*1;
WAR1 BY W1 - W3*1;
PRACWAR | PRAC1 XWITH WAR1*.5;
PRACCON | PRAC1 XWITH CON1*.5;
CONWAR | CON1 XWITH WAR1*.5;
PRACCONWAR | CON1 XWITH PRACWAR*.5;
PRAC1@1; WAR1@1; CON1@1;
PA1 ON PRAC1*.5;
PA1 ON CON1*.5;
PA1 ON WAR1*.5;
PA1 ON PRACWAR*.5;
PA1 ON CONWAR*.5;
PA1 ON PRACCON*.5;
PA1 ON PRACCONWAR*.5;
 Kelsey Borner posted on Monday, May 11, 2015 - 3:38 pm
I could not fit the additional information into my previous post:

I am trying to run a monte carlo simulation study to estimate the sample size needed to conduct a three-way interaction. I received the fatal error message at the beginning of my message, and am unsure of what to do. Thank you so much for any help!
 Bengt O. Muthen posted on Monday, May 11, 2015 - 6:38 pm