ApolloChoiceModelling forum

Dear Concern,

I have 6 latent variables, in which the measurement indicators (likert scale 1-5) of the 2 latent variables are not answered (in some cases) by the respondents and as such, the CSV file has NaN'S field. I tried to fill them with 0. but the problem persists in the HCM model with OL says that:

Some levels in 'outcomes ordered' do not exist in 'coding' for the model component.

Is there any way to deal with missing value in order to run the HCM+OL model?

Hi

apologies for the slow reply.

Do you want to exclude those people from that part of the likelihood? That would be quite standard, and can be achieved by the "rows" setting, so e.g. rows!=0 if you want to exclude the ones at 0

Stephane