Latent Class Analysis (LCA)

Latent Class Analysis is a cluster-wise regression approach that we use to discover respondent segments with similar (latent) preference structures in choice data. Different from traditional cluster techniques, it first classifies respondents into segments that are as distinct as possible, and then it estimates preference structure parameters at the segment level instead of the respondent level.

Latent Class: benefits & limitations

• It delivers a good understanding of market structure and of addressable target groups
• It delivers insightful input into product portfolio optimization by providing insight into heterogeneity in preferences
• It delivers a perfect balance between insightful but potentially unstable respondent-level parameters and solid but over-generalizing sample-level parameters

Latent Class: when to use it?

• Perfect input for product line optimization
• To identify unmet needs for product development
• To identify different price groups and what each group values, so they can be targeted