www.cmm.bristol.ac.uk/forum

Hi it's me again. I need to run different models for different types of outcomes of social support, and now I tried a Poisson model and it went fine, but the Neg binomial is constantly crashing.

This is the formula.
#######################################################
m8 = log(AltSupportCount) ~ 1 + TieIsFamily + TieSameSex + TieSameAge + EgoMarital + EgoSex +
EgoTMig + AltSex + AltIsRoma + AltNatRomanian + AltResidence + NetPropFemCen + NetPropFamCen +
NetPropIsRomaCen + NetPropNatRomanianCen + NetPropResidCen + EgoTMig:TieSameSex + EgoSex:AltIsRoma +
EgoTMig:AltIsRoma + (1 | EgoIdentifier)
(VarCompModel <- runMLwiN(Formula = m8, D = "Negbinom", data = RomaAllData,
estoptions = list(EstM = 1)))

This is the error on the console
"error while obeying batch file /var/folders/3n/zvw9lr5s1c1dnl0jkrc8y2xh0000gn/T//Rtmp1LfSTP/macrofile_9f41447b133.txt at line number 195:
MCMC 0 500 1 5.8 50 10 C2498] C2499] 1 1 1 1 1 8
MCMC Error 0010 : Wrong parameters for Gamma distribution generator
Execution completed"

Best,

It's hard to know what is going wrong here without a reproducible example. My suggestion would be to check the IGLS estimates (i.e. remove EstM=1) for the model and see whether they look sensible, as these will be used as starting values for MCMC. I would also suggest checking your response variable to ensure that all values are positive.