Max-Diff Mixture Models

From Survey Analysis
Jump to: navigation, search

The main mixture models used to analyze the Max-Diff experiments are:

  • Latent class logit models, which assume that the population contains a number of segments (e.g., a segment wanting low priced phones with few features and another segment willing to pay a premium for more features) and identifies the segments automatically.
  • Random parameters logit models, which assume that the distribution of the parameters in the population is described by a multivariate normal distribution. This model is sometimes referred to in market research as Hierarchical Bayes, although this is a misnomer. See Tricked Random Parameters Logit Model for an example.
  • C-Factor models, which can be either latent class or random parameters logit models, but additionally allow for heterogeneity in scale factors. See the Latent GOLD website for examples.


Sawtooth Software has modules for estimating latent class logit and random parameters logit models with max-diff data using the Tricked Logit Models.

Latent GOLD Choice can estimate C-Factor latent class models using both a tricked logit model and other variants of standard logit models.

Q has routines for estimating latent class and random parameter logit models using Rank-Ordered Logit Models With Ties.

A more up-to-date version of this content is on