Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Multi-variate Latent Class?

I recently ran a study incorporating two MaxDiff exercises – one regarding Needs and one regarding Uniqueness, both on the same set of items. I wanted to check in with your team to see if it’s possible to run a Latent Class analysis with both exercises included? Of course, this would help us to better identify if there are any differences in preferences across both dimensions concurrently – versus siloed within each.
asked Apr 11 by Alison

1 Answer

0 votes
If you combine the two data files into a single file (e.g. export the two .cho files and concatenate them, making sure to group together  the observations for a given respondent) then the LC analysis would run and would help you group respondents.  I might want to think it through first, but if you want to take into account that your two DVs are different, you might want to pretend for the purposes of this combined model that you have twice as many predictors as you do - the attributes as they predict uniqueness and then the same attributes as they predict needs.
answered Apr 12 by Keith Chrzan Platinum Sawtooth Software, Inc. (79,225 points)
...