Part 4: The Basics of the Affinity Sort
The Bricolage of Discovery: Synthesizing Innovation Insight Across Disciplines and Studies
I started my journey in qualitative data analysis with hundreds of dog-eared printing press new owner registration cards. In addition to contact information the cards had a comment section and my job was to figure out what the customers were saying. I read them. And after an extraordinary struggle with the handwriting I still had no idea where to begin. I piled similar comments together. I put a label on each pile. I eventually figured out meaningful categories of comment and then probably counted how many I had in each pile (oops, big mistake). Over time I developed a few guidelines.
- Data preparation is worth the time
- No two qualitative data points are exactly the same. (Even if the words or images are very similar the metrics of things like “source” can be different and meaningful (male-female; age; region, etc.).
- An observation can be sorted into one, or multiple piles
- All observations may be important, but not all observations have meaning for your charter
- Repeated observations are not made more important by repetition (numbers don’t count; 1 and 25 have the same “nominal” value)
- Label your piles … but don’t rush to make the labels have meaning too quickly… give this some hang time.
My clustering skills were refined in a search for meaning in boxes of studies (most in raw data form) on an outbreak of Multiple Sclerosis in the Faroe Islands after WWII. The data had been gathered for over 25 years, by a variety of researchers, in a variety of ways, in multiple languages. What we are doing is creating the raw conceptual material for the description of a unique space. We are seeking to perfectly channel the voice of the customer with each decision we make.
In the next issue I will describe the process for finding meaning and perhaps understanding in your piles of data.
The affinity sort is part of an overall process being developed for a book to be published in January 2017. Chapter process components will include:
- The nature of data
- Structuring diverse nominal data for analysis
- The importance of bias vs. informed intuition
- The basics of an affinity sort,
- Converting affinity buckets into qualitative clusters with definition (meaning in the bricolage)
- Selecting vs. eliminating
- Moving from qualitative clusters to a testable hypothesis
- Qualitative testing and future myth building
- What is your communicable truth
- Application to discovery and informing the charter
Chris would enjoy hearing about your synthesis experience, the good, the bad and the ugly… and as always volunteer readers are welcome.