Written by: Jay Magidson, Statistical Innovations Inc. and Gary Bennett, The Stats People Ltd.
Jay Magidson gave the following presentation at the 2016 Advanced Research Techniques Forum of the American Marketing Association (AMA) on Monday 27 June.
This presentation describes a new approach on how to develop a MaxDiff typing tool and uses real data to illustrate our approach.
Marketers often find it valuable to include a relatively small number of “Golden Questions” in future surveys in order to classify respondents into identified segments with different needs. Specifically, this method proposes inclusion of a relatively small number of binary items in future surveys, whereby each item requests a simple choice between a specified attribute pair, so providing a way for selecting such items and the resulting classifications.
Results suggest that based on these binary items, a high percentage of the original respondents could be expected to be assigned to the correct segments. The resulting typing tool is ultimately simpler and an improvement over earlier approaches as it allows for customisation. By tailoring the model in this way, a higher accuracy is achieved for those segments judged most important to target.