Publication date: 08/13/2020


Fit Models for MaxDiff Experiments

Use MaxDiff (maximum difference scaling) as an alternative to standard preference scales to determine the relative importance of items being rated. MaxDiff forces respondents to report their most and least preferred options. This often results in rankings that are more definitive than rankings obtained using standard preference scales.

The MaxDiff platform enables you to do the following:

Use information about respondent (subject) traits as well as product attributes.

Integrate data from one, two, or three sources.

Obtain subject-level scores for segmenting or clustering your data.

Estimate subject-specific coefficients using a Bayesian approach.

Use bias-corrected maximum likelihood estimators (Firth 1993).

Figure 5.1 MaxDiff All Comparisons Report 


Overview of the MaxDiff Modeling Platform

Examples of the MaxDiff Platform

One Table Format
Multiple Table Format

Launch the MaxDiff Platform

Launch Window for One Table, Stacked
Launch Window for Multiple Tables, Cross-Referenced

MaxDiff Model Report

Effect Summary
MaxDiff Results
Parameter Estimates
Bayesian Parameter Estimates
Likelihood Ratio Tests

MaxDiff Platform Options

Comparisons Report
Save Bayes Chain
Want more information? Have questions? Get answers in the JMP User Community (