Categorization analysis, also known as discrete analysis proceeds by:
- Categorizing each person's data for each feature
- Creating a summary table of the categorizations
- Data cleaning
- Assigning each attribute to a category
Categorizing each person's data for each feature
There are 25 possible evaluations that a person can provide for a feature. The table below shows how these 25 evaluations are categorized (there are multiple slightly different variants of this table). For example, if a person says they Like it when the feature is available and they Dislike it when it is unavailable, it is categorized as Performance.
Creating a summary table of the categorizations
Once the categorizations have been computed, the next step is to create a summary table showing all of the data, as illustrated below.
After creating the summary table it can be useful to perform data cleaning. See Data Cleaning for the Kano Model.
Assigning each attribute to a category
A standard approach to analyzing the Kano Model seeks to assign each attribute to one and only one category. The right-most column below assigns each attribute to the highest category.
In practice, it can be difficult to assign an attribute to a specific category. For example, looking at Heart Monitor in the first row, is it Indifferent or Attractive? Some practitioners calculate the difference between the highest and second-highest category, which is called category strength, and only assign a category if the highest category is at least five percent above the second-highest (i.e., strength is at least 5%). This approach is highly dubious, as the small sample sizes of most Kano studies mean that such differences will not be statistically significant.
The table below uses column comparisons to show statistical significance. Only two of the attributes - Glucose monitor and Two day battery life have sufficient category strength that we can even say they are significantly different from all the other categories.
Where it is not possible to clearly assign attributes to categories, it becomes appropriate to instead focus on relativities, as is done with Satisfaction Coefficients Analysis of the Kano Model and Continuous Scale Analysis of the Kano Model.