Previous Topic Next Topic
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view


I am running some latent class models with civariates using poLCA in R, and
have a question the meaning of some error messages.

My model has 8 y-variables (dichotomous) and 2 x-variables (also dichotomous),
and I am trialling different numbers of classes.

When I run a 2-class model, the model runs fine without any error messages. I
used the nrep = 10 option to avoid local maxima. Class % are ~81% and 19%.

When I trial a 3-class model,  firstly I upped the nrep to 200, since for some
runs it isn't till run 139 that the maximum is found. Class % are ~81%, 16%
and 2%. These vary a LOT if a model based on a local maximum is used.
Anyway, for the final model with max, I get the following message:

Warning messages:
1: In sqrt(diag(VCE.beta)) : NaNs produced
2: In sqrt(diag(VCE.beta)) : NaNs produced
3: In sqrt(diag(VCE.mix)) : NaNs produced
4: In sqrt(diag(VCE.beta)) : NaNs produced
5: In sqrt(diag(VCE.mix)) : NaNs produced

What does this mean for the model?

[hidden email] mailing list
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
Reply | Threaded
Open this post in threaded view

Re: poLCA

I understand that the NaN's mean that something can't be computed, I'm just unsure as to what they refer to. Can anyone shed some light as to what the beta and mix matrices are?