The Affinity Sort: A way to gain a fresh perspective from the same old data
In the 1970’s I worked for Ditto as the Market Research guy. We made little printing machines that were popular in schools for 75 years until xerography killed off the technology. We would get comment cards back on each machine we sent out.(Teachers are very good about following directions and filling out a short questionnaire and the comment card was one of the set up instructions.) The hand writing was so good on these cards that we never bothered to transcribe them. We would sort them into piles. We would give the piles names and draw conclusions. We could do it all over a month later with a different mind set, a different question, and get a different answer from the same cards. This simple method is called an “affinity sort”. 40 years later I find myself starting every analysis of nominal data with the same old process. We have added a few interpretive exercises to that method, but it remains the basis and beginning of all nominal data analysis.
The term “Affinity Sort” represents a large set of post research tools which are designed to help us interpret the data. Essentially it means “putting like things together”. Clearly what is alike depends on two key elements:
- The field research collected
- Quotes
- Photographs
- Behavioral observations
- Results from multiple studies using different methods
- The perspective of the interpreter
- A priori hypotheses
- Emergent hypotheses
- Biases
- Strategic mindset
The tools available to us today are much stronger than the ones we had 30 or 40 years ago. I am not always sure that the results are better. (I remember one of those Ditto cards with a skull and cross bones drawn beside the question…hard to code that data, but the meaning was clear to us.) The primary power in any affinity sort is the sorters familiarity with the data and the willingness to take the time to sift, discuss, and sift some more.
How to Conduct an Affinity Sort
To prepare for the exercise individual team members or single interview teams should review all of the observations from their notes. On a pad of post-it notes, cards, or paper put a single observation, picture, artifact or item on each card. When you are finished you should have all of your quotes, raw facts, and observations on a stack of notes that can be sorted.
The full team (all of the people in the field and anyone else that can be an equitable resource) begin by gathering all the cards together. As a team:
- Put the first observation on the wall.
- Look at the second observation – is it the same or different
- Same: add it underneath the first observation to create a “bucket”
- Different: start a new bucket (put that observation next to the first one on the wall)
- Meaningless/ can’t understand: set aside
- Note that one observation may fit in more than one bucket
- Look at the third observation. Does it go in the same bucket as either of the two observations? If not, create another bucket.
- As piles start to form give them useful names for the research task, in other words, title the bucket. Write the name on a different color post-it note and put it at the top of that bucket on the wall.This can be done during the sort; it is not necessary to wait until the sort is over.
When a consistent pattern seems to have emerged break into teams of 2 or 3 (protecting team diversity where possible) and as a breakout team, continue the process until all observations have been sorted.
Quickly revisit the set-aside pile to see if there are unique items that now fit and then review piles:
- Very large piles should be considered for breaking into several different piles
- Very small piles should be considered for collapse with others
- All piles should have meaningful labels or titles
How to do a Qualitative Cluster Analysis
The following steps are the Qualitative Cluster Analysis™ (QCA) process which is a tool that can be used to sort and discover meaning in any set of nominal data from a single study or across studies. It is a useful technique for opening up the data for interpretation.
Process Steps
- Affinity sort – the simple process of putting like things together labeling the things that are alike with a title meaningful in terms of the research task
- Nominate a central theme: This can be a single bucket title from the affinity sort or a combined theme from several different buckets.
- Identify secondary and tertiary themes: Support with affinity buckets and show and label relationships (below connections were symbolized by string and representative statements were posted to illustrate themes and areas of opportunity).
- Give the QCA a title
- Discuss for new learnings and determine the most appropriate way to communicate this with others
Moving from Clusters to Themes
A theme is something that you have discovered that is meaningful to you as a way of understanding the customer experience. Take the responsibility to translate the theme with its customer voice into an appropriate format with language for your internal customer E.g. bulleted notes, story, story boards, etc.
Analogical Modeling
A metaphor is an analogy and analogical modeling can be used to open up the data. It uses comparison and, in this case, provided by analogy to help tease out what’s going on in the data.
- Select a metaphor e.g. an automobile, coral reef, a movie – something that you know enough about the component parts and their interdependencies
- Deconstruct the metaphor – it’s component parts and interdependencies
- Review the data and align data with the metaphor
- Review the work and generate insights, opportunities for the research task
- Report out in appropriate format
Theme Posters
Take the themes from the Qualitative Cluster Analysis and as a team decide where there are opportunity areas for each cluster. Then break into teams and each team take a cluster and an opportunity area and develop a
theme poster depicting the problem and opportunities.
Storytelling
A story can be a compelling way to convey a need and can make the listeners of the story advocates for the:
- Build a story about a happy ending for the customer
- “It was a dark and stormy night…”
- Pick a respondent or build a persona that represents a segment of your respondents and feature them as the main character
- Build in learnings from QCA
- Allow the story help explain how the customer thinks, decides, feels delighted
- Do not limit yourself to today’s reality as you visualize the positive end game.
A story modified from a Coleman team’s work
Jim pulled into his garage at close to 7:00 PM. He was tired and still felt he had a lot of work to do. As the garage door went down the whole property security screen went back up. Jim remembered when he had had to carry a house key and punch in codes. He walked through the kitchen, grabbing the newspaper on the way, and out onto his deck. It was only a little warmer outside than it had been in his house or car. He knew that the temperature was still nearly 90 and the humidity almost as high, but his amazing outdoor climate control gave him the feeling of a cool June evening not the reality of August in Chicago.
He pulled a beer out of the deck cooler and sat down in his favorite chair. Immediately he spotted a Red-Winged Blackbird sitting on the reeds growing in his little swamp. The swallows swooped and dived overhead. In the back of his mind he knew it was the huge mosquito population that drew these graceful creatures to his yard but it had been years since he’d been bitten, thanks to his bug-away system.
The phone rang. It was his wife Jane. She had just picked up the kids on her way home from the office and asked if he’d mind cooking tonight. Jim said “sure” as he picked up the remote. He hit his pre-programmed “super dad” meal. The grill started to heat and the burgers went into a fast thaw cycle as he read the paper.
Snapshots from the Field
These profiles are completed in the field. Look at each profile and determine the problems, challenges, and needs that each respondent encounters while engaging in the activity or using the product you are researching. As a team, come up with concepts to solve the problems of one respondent at a time. Afterward choose the problem areas and concepts that could be representative of a target market for further development.
