R help - Nov 2010 - Mixed multinomial logit model (mlogit script)

Dear all,

I am trying to run a mixed multinomial logit model in R since my response
variable has 4 non-ordinal categories. I am using the package mlogit that
estimates the parameters by maximum likelihood methods. First of all, I prepared
my data using the mlogit.data command. In the mlogit command, one can introduce
alternative-specific (fixed factors??) and individual-specific (random
factors??) variables. Each of them is indicated before and after the | ,
respectively, in the formula script. My model has both types of predictor
variables. As the author of the script indicated in the mlogit helps, I am using
the mixed model script example:

m<-mlogit (mode ~ price + catch | income, data = Fish)

but I obtain the next error:

Error en drop(.Call("La_dgesv", a, as.matrix(b), tol, PACKAGE =
"base")) :
  rutina Lapack dgesv: sistema es exactamente singular

The only syntax that I can run is:

m<-mlogit (mode ~ 1 | income, data = Fish)

which I don not understant really well what is going on but I understand that
here just individual-specific variables are being analyzed.

So, I am a beginner pearson in R as well as in this kind of models. I would be
really grateful any help.

Thanks in advance,

Maria.




      
	[[alternative HTML version deleted]]