What is a Max-Diff Technique?
Read TRC White Paper: How to Improve Your Segmentation with Max-Diff
Max-Diff (Maximum Difference Scaling) Technique Identifies Most and the Least Critical Factors (Attributes)
Having respondents rate the importance of various factors is not the best way to identify those that are most valued. Follow an importance-rating-centric approach and you'll usually get an 'everything is critical' distribution (see chart A for an example of results).
It's better to have people make choices, and Max-Diff is a powerful method for doing that. Max-Diff combines elements of pairwise comparisons with the design power of discrete-choice to deliver (a) a hierarchy of attribute importance and (b) a sense of the distance between ranks.
Respondents are shown a set of four to five attributes, and asked to tell us which among them is most important and which is least important. This task is then repeated several more times with a systematically chosen series of attribute groupings. The output from this exercise provides a strong sense of attribute hierarchy at the respondent level, which can in turn be aggregated to show how particular populations prioritize their decision-making (see chart B for an example of how such a hierarchy might play out).
Max-diff can handle a maximum of about 20 attributes. After about 20 items, the task can become tedious so another recommended alternative is an adaptive approach such as TRC's Bracket™. For instances when results are needed fast and on the budget a streamlined version of Bracket called Message Test Express™ (MTE™) is a a proven solution (see our case study).