choice-metrics.com

I want to analyse my DCE data. I have used block stratification, where two blocks were used.
For the analyses would you prefer to combine the scenarios of the two blocks and analyse it within on model.
Or would it be better to make two separate models for the two blocks, or maybe three models. Or would this have an impact on the results?

You could ideally estimate a joint model because this gives you more data points and increases statistical power, including comparing parameters between the blocks.

You could first estimate two separate models, and observe in which parameters the differences across across the blocks. Then you can estimate a joint model and decide to make the parameters that are similar generic while others are block-specific.

Michiel