Create new variables which contain the individual-level coefficients from a MaxDiff latent class analysis, MaxDiff hierarchical Bayes or MaxDiff ensemble output. You should select the output before running this script or in the MaxDiff's object inspector click on SAVE VARIABLE(S) > Proportion of Correct Predictions.
Example
When ran from the menu the script produces both a variable set of the zero-centered utilities for the parameters in the model for each respondent as well as a SUMMARY table:
Technical details
Prediction accuracy is measured in terms of the proportion of a respondent's questions in which the best alternative was correctly predicted. If cross-validation is used, the predictions are made on only the holdout (out-of-sample) questions. Otherwise, predictions are made on all of the respondents' questions.