ApolloChoiceModelling forum

Hi, Prof. Hess,
I encountered a problem in estimating the ICLV model. We all know that sometimes, not all values in attitudinal indicators are selected. For example, in my study, no one chooses 1 in the indicator variable “CA1”. Then, the model reported an error message:

Some levels in "coding" do not exist in "outcomeOrdered" of model component "indic_CA1"!

How can I solve this problem?

Best,
Yue

Hi

if a level is never selected, then you cannot estimate the same number of thresholds.

So for example, if your question is on a scale from 1-5, you would normally estimate 4 thresholds. But if level 1 is never chosen, then you can only estimate three thresholds.

The easiest way of dealing with this is to reformulate your dependent variable, for example:
In your case, 1 and 2 would then be code as 1, 3 as 2, 4 as 3, etc

Stephane

Thanks for the prompt reply prof Hess! I have solved the problem with your help. Another question is that after the estimation process, only the coefficients of the variables are given, and other results always being calculating. Could this be related to the hardware of the computer?

Hi,

Could you be a bit more specific about what other results are not showing? Could you share your code and point where the problems are?

Cheers
David