I want to contruct a quadratic growth model with ZIP variables. Have I got the model specified correctly? If so, do you have any suggestions for how I can avoid the fatal error of insufficient memory?
TITLE: SAFETALK transmission risk perception longitudinal analysis - no mediation quadratic growth model with zero inflated poisson vars, with uavi < 130
DATA: FILE IS safetalk063010LongMed.dat;
VARIABLE: NAMES ARE arm numUAVI1 numUAVI2 numUAVI3 numUAVI4; USEVARIABLES ARE arm numUAVI1 numUAVI2 numUAVI3 numUAVI4; USEOBSERVATIONS ARE (numUAVI1 < 130) and (numUAVI2 < 130) and (numUAVI3 < 130) and (numUAVI4 < 130); COUNT ARE numUAVI1-numUAVI4 (i); MISSING ARE ALL (999);
I would suggest using INTEGRATION = 7; in the ANALYSIS command and asking for SVALUES in the OUTPUT command. Once you have the starting values, you can rerun with the default INTEGRATION = 15; to check for stability.