MaxDiff scaling (aka Best-Worst scaling) has become a widely used research technique because it provides much better discrimination than straight rating scales applied to linear attribute lists.

Instead, MaxDiff tradeoff designs pit smaller random subsets of attributes against each other: which attribute in the subset is most important (or appealing), which one is least important (or appealing). A more focused, simpler task for respondents than rating a lengthy attribute list on importance.

A Potential Limitation

But until recently, MaxDiff approaches often limited the number of attributes that could be realistically accommodated in a study. That’s because MaxDiff designs typically require each respondent to complete a large number of tradeoff sets within a given survey. The more attributes, the more tradeoff sets, lengthening the survey and displacing other needed metrics.

The usual solution: reduce the number of attributes so that the MaxDiff module will “fit” comfortably within the study framework.

The down-side: a loss in granularity and the opportunity for insights that may have been surfaced with a full-ranging, uncompromised attribute list. Depending on the research objectives, this could be a particularly important consideration.

Improved MaxDiff Designs

The good news is there now are state-of-the-art, improved MaxDiff design options available that our Marketing Science team has been trained on and we can recommend. Several of these approaches embody hybrid, advanced modeling that overcome content limitations and strengthen attribute discrimination. More good news: while the modeling is complex, the usage is easy and the applications many.

These new alternatives to traditional metrics can be of value when objectives call for the evaluation of large number of statements/variables such as when screening a wide range of current and alternative flavor extension ideas or brand claims.

The improved MaxDiff designs are also worth considering when there is a desire to reach beyond identifying “winners and losers” to glean further insights and potential growth opportunities, for example when determining brand benefits/attributes hierarchy for equity extension and the underlying benefit/attribute structure of the equity.

Ultimately the choice based approach we recommend will depend on:

  • Business and learning objectives
  • Scope of the research: Is MaxDiff tradeoff the primary thrust or a smaller section within a more comprehensive, multi-faceted study?
  • The number and complexity of the items to be included in the MaxDiff.