Q offers a number of different ways to access Latent Class regressions. Here are some of the methods and when you should use them.
There are three menu-based ways of running Latent Class Regression in Q:
- The most general method is using Create > Regression > Mixture Models which will, if left at its default settings, perform Latent Class Regression. The data needs to be set up as an Experiment question (see Experiments Specifications).
- Create > Regression > Latent Class Analysis also performs Latent Class Regression, but permits the user to select non-regression data as well (e.g., attitudes). The data needs to be set up as an Experiment question (see Experiments Specifications).
- Create > Marketing > MaxDiff > Latent Class Analysis. This will give the same results as the other two approaches however:
- it is easier to use, as the data setup is more straightforward.
- The first two approaches have more diagnostics for selecting segments, will automatically search through different numbers of segments and different random start points. In general, they are better if the focus is on segmentation.
- The MaxDiff-specific methods produces cross-validation statistics, which is better if the goal is to make conclusions about the validity of the segments or compare them to HB analyses.