This article describes how to go from a MaxDiff Latent Class Analysis output:
To new variables which contain the Sawtooth-style preference shares (formula here) for the alternatives in a MaxDiff latent class analysis, MaxDiff hierarchical Bayes or MaxDiff ensemble output. The shares are computed from the individual-level coefficients generated by the MaxDiff analysis.
- A document containing a MaxDiff Latent Class Analysis output.
To create variables that contain estimated preference shares for each respondent, based on the latent class analysis:
1. Select your latent class analysis output.
2. Select Anything > Advanced Analysis > MaxDiff > Save Variable(s) > Sawtooth-Style Preference Shares (K Alternatives).
When shown in a SUMMARY table, the output will show a summary table of the new variable.