ApolloChoiceModelling forum

sorry, not at present. It should be possible in a month or so

hi

this was in an earlier version of Apollo, and is not currently implemented. It will be added again in the near future, but requires a change to the bgw optimiser first

Stephane

Hi

yes, these differences are very large, and also go in the opposite direction of what we typically see. Did you try a basic version of the model, without all the attributes?

Stephane

Hi

apologies for the slow reply. Could you share your code and data with me outside the forum and I'll have a look for you

Stephane

hi

what I mean is that you could estimate a parameter for each level, and then plot their estimates against the values and see if it looks linear

Stephane

Hi

in general, using LVs in the class allocation model is empirically difficult and often leads to issues.

In your case, I would look at those parameters for which you do not obtain classical standard errors

Stephane

Hi

no, it cannot. The weights need to be provided by the user as values

Stephane

Hi

in your case, it's likely the line breaks. See the point at https://apollochoicemodelling.com/faq.html

Search for Why do I get an error saying that one of my parameters does not influence the likelihood, even though I am using it in a utility function?

Stephane

That's simply ensuring that the weights sum to N, but the same issues remain. It's much better to just incorporate interactions, and then reweight your results

Hi

difficult to know without looking at the data. can you share the data and code with me outside the forum and I'll have a look

Stephane