Over the past few years, MD Analytics has encouraged the use of new survey methods in healthcare and pharma marketing research that more accurately measure physicians’ attitudes and opinions in regards to treatments for various conditions. For instance, we favour the utilization of MaxDiff (a contraction of “Maximum Difference Scaling”) rather than more traditional rating scales, whenever possible.
How important is all this, anyway?
MaxDiff has been shown to provide greater discrimination between attributes, thereby providing greater value to the end user. However, one weakness of MaxDiff is that its measure of importance for one attribute is made in relation to the other attributes, but not against an objective standard. In other words, attribute A may be twice more important than attribute B, but it could very well be that neither attribute A or B are important at all.
Introduction of Anchored MaxDiff
This lack of information can now be addressed by adding an anchoring method to the MaxDiff exercise. Anchored MaxDiff allows us to draw a line in the sand, and determine whether attributes pass that threshold or not. This anchoring method consists of an additional question that asks which attributes presented, if any, are “must-haves” for the respondent. Those attributes selected as “must-haves” would pass the threshold, while those unselected would not.
The result of an anchored MaxDiff is somewhat different from a regular unanchored MaxDiff. Whereas a regular unanchored MaxDiff presents a “share” of importance (in which, for example, item 1 is twice more important than item 6), the result from anchored MaxDiff is an index on which the threshold is set to 100 (or any other number of choice). Attributes deemed important are greater than 100, and attributes deemed unimportant are lower than 100. The results keep their ratio properties, which means that we can still say that item 1 is twice more important than item 6. It should be noted that because the anchoring method is a supplement, rather than an alternative to MaxDiff, results can be presented either way.
So the attribute that matters most to me also matters the most to physicians… or does it?
Latent Class Segmentation in MaxDiff
Another new capability with MaxDiff is the ability to create a needs-based segmentation solution based on respondents’ answers to the MaxDiff exercise. Using the example described above, let’s assume that “item 8” is an attribute that we were most interested in physicians’ reaction. It may come as a disappointment to find “item 8” ranked so low in relation to the other attributes measured.
However, running a segmentation analysis on the MaxDiff data reveals that one-half of physicians consider the “item 8” feature to be among the most important aspects they look for in a treatment, while the other half of physicians largely ignore it. Other data collected from the rest of the questionnaire would then provide further information on the profile of this segment of physicians, and determine their usage and attitudes with current treatments.
To learn more about our Anchored MaxDiff or Latent Class Segmentation with MaxDiff and how it can be applied to your marketing research objective, please contact us.