Scaling First-Last, MaxDiff and Best-Worst Data
Ennis, D. M. (2009). IFPress, 12(3) 2-3.
Among market research practitioners, there has recently been interest in scaling product or category characteristics (such as possible benefits) based on responses indicating the items with the greatest and least magnitude among a subset of possible items. The basis for this choice could be liking, purchase interest, importance or even a sensory characteristic such as sweetness. For example, considering the characteristics of plug-in air care products, items to consider might include “low cost,” “does not fade over time,” “has a use-up cue,” and “has a fresh scent.” A respondent may be instructed to choose the attribute of most importance and the one of least importance in making a purchase decision. From a large collection of items, subsets of equal size are chosen and presented in a balanced design. The typical number of items used per respondent is four. The analytic task is to develop a scale on which each attribute can be placed so that scale values for all of the items from most to least can be obtained.
Figure 1. Typical responses in a first-last choice task.