How to Run a Boosted Varying Coefficients Model on MaxDiff Data

The table below shows the output of a boosted varying coefficients model using MaxDiff data on technology companies. In this case, 2 covariates have been selected (likelihood of recommending Apple and Samsung), and a boosted 3-class latent class analysis over respondents is run at the end. The histograms show the distribution of preference shares across respondents.

Requirements

A document containing your MaxDiff respondent data.
A MaxDiff experimental design.

Method

Add the latent class analysis to your project from the Report tree by selecting + > Advanced Analysis > MaxDiff > Multinomial Logit.
Select your experimental design. You can use an existing table, an R output, variables from a data set, or a URL. For this example, I'm using an existing table that I've added to my project by copying it from Excel and pasting it onto my page using Ctrl + v (or Cmd + v on a Mac).
The Design input table should be in a format similar to the one shown below. The 'Version' column is optional when there is only one version in the design. The 'Question' column is also optional. However, if these columns are included, they must be named 'Version' and 'Question'. The columns after this contain the indices of the alternatives presented to the respondents.
OPTIONAL: Click on Add alternative labels, and enter the alternative names in the first column of the spreadsheet editor. The order of the alternatives should match the order in the design.
In the Respondent data section, select the Version variable from your respondent data set. If you have only one version, leave this blank.
In Best selections, choose the variables in your data set that identify the options that were selected as best, or most preferred, for each task. The order of the variables you have selected should match the order from the design (i.e., the variable for the first task should be selected first, the variable from the second task should be selected next, and so on).
In Worst selections, choose the variables in your data set that identify the options that were selected as worst, or least preferred, in each task.
In the Model section, update the Type to Varying Coefficients.
Select the Covariates.
OPTIONAL: tick Additional latent class analysis to run a final latent class analysis over respondents.
Choose the Number of classes. I've selected 3 classes for this example.
OPTIONAL: In the MaxDiff logit dropdown, choose between Tricked Logit and Rank-Ordered Logit with Ties. The former is faster, but the latter is used in latent class analysis for MaxDiff.
Set the number of questions left out for cross-validation, which is the number of questions to leave out per respondent for cross-validation. The default is 0.
Set the Seed, which is the random seed used by the model and also used to determine questions to leave out for cross-validation. The default is 123.
Click the Calculate button to run the model.
OPTIONAL: In the Save Variable(s) section, choose the variables you want to save. Choices are Class Membership, Class Membership Probabilities, Preference Shares, Proportion of Correct Predictions, RLH (Root Likelihood), Sawtooth-Style Preference Shares (K Alternatives), and Zero-Centered Utilities.
OPTIONAL: From the Report tree select + > Advanced Analysis > MaxDiff > Diagnostic to request the following diagnostics: Class Parameters Table, Class Preference Shares Table, Parameter Statistics Table, Posterior Intervals Plot, and Trace Plots