MaxDiff Scaling

Measuring the importance and preference of benefits is often done with common methods such as rating, ranking or chip allocation. However, these methods often lead to non-discriminating or heterogeneous answers. Usage of Maximum Difference Scaling leads to more discriminating ratings and more refined results

MaxDiff is able to cope with many items. Furthermore, the data are scale-free and not biased by cultural differences.
With Maximum Difference Scaling the respondent is asked to reveal his/her most and least attractive benefit. This exercise will be repeated a number of times for a changing subset of benefits. The tasks are usually quite simple to answer and are engaging.

The nature of the technique and the form of the data make a combination of MaxDiff experiment and Latent Class estimation and excellent pair to investigate key preference segments.
"Under water" MaxDiff uses experiment design techniques similar to CBC as well as hierarchical Bayes techniques to estimate individual preferences.

Software for MaxDiff experiment design, scripting and analysis is available. Please read our Software pages for more information.

Very suitable for: claim testing, message evaluation.