Affinity Diagramming


Affinity diagramming is a conceptual organizational exercise that can have numerous benefits for problem solving in team environments. Ideas are not only identified by a group, but then explicitly categorized and grouped. Enumerating instances and aspects of cultural phenomena—and categorically naming them—identifies trends and highlights differences. A phenomenon, whether it is a behavior or a problem or both, is best understood when the interlocking parts of the system can be made explicit.

History and Theory

Affinity diagramming was developed by Jiro Kawakita over the course of fifteen years while he was conducting ethnographic research in the Himalayas. Its popularity in Japanese management culture spread primarily through its introduction at the Free Campus University, at which Kawakita was a professor (Scupin 233-234). The KJ Method is the application of affinity diagramming, or the formalization of verbal/written procedure. Kawakita’s proposition suggests a connection between unstructured observations and scientific hypothesis testing or experimental design, which heretofore was arcane and non-specific (10).


“An affinity is built from the bottom up by first grouping similar observations, labeling them, then building larger groups out of these small groups” (Beyer 30).  Team members—ideally a group of several—will follow three steps to create the affinity, and additional step of analysis or reflection. Scupin describes the steps as label making, label grouping, chart making, and analysis (235). The following descriptions for each step are an amalgam of the KJ Method as described by Hoerl & Snee, Scupin, Kawakita himself, as well as experienced by a tutorial given by Professor Kristian Kloeckl at Northeastern University in October of 2016.

I. Label Making

Assemble a group of several participants. Hand out pieces of paper, big enough to write a sentence or phrase on, to each participant. Post-It® notes work as well. Clear a board or wall or a table to act as a canvas. Label the center of the canvas with the problem/phenomena concisely. Allow each participant to brainstorm ideas, concepts, objects, actors, interactions, behaviors, emotions, or related phenomena associated with the target problem/phenomena. As a participant thinks of an idea, they write it on an individual piece of paper, verbalize it to the group, and adhere it to the canvas in no specific place. Do such until the enumeration begins to slow down.

II. Label Grouping

Then, the team begins to group the ideas on the canvas by whichever criteria seems rational. The team does this simultaneously and silently. Participants are encouraged to group, regroup, split and combine papers on the canvas despite another participants opposing groupings. Once a distinct amount of groups emerges—or participant consensus or attention fatigue sets in—the grouping phase is over.

III. Chart Making

Once the groups of paper have been established, participants begin to devise titles or categorical names for the groups of labels. In discussion of titles, individuals can propose titles that restructure, or oppose the groups of labels made in the previous step. If any phrase or concept is perceived as an outlier, it is helpful to have the author explain the experience that led them to include this. Often these clarifications lead to the outlier being fed into an existing group, or expanding another grouping’s title. Once every grouping of concepts has a title, larger structural questions should be asked: do certain groupings fall completely within others? and do certain groupings share common traits with other label groups? Arrange the chart utilizing structural visual metaphors, such as inclusion/subset, opposition, or union.

IV. Analysis

No affinity diagram is complete without analysis. Kawakita recommends an analysis that is both concise and smooth (12).  A verbal and or written explanation guides the team in distinguishing the interpretations from the descriptions. It reduces the complexity of the enumerated labels into a form that is manageable and consumable by parallel participants, or amateurs, or clients (Scupin 236). The analysis can begin to explore what labels are causes and which labels are effects. Additionally, the analysis can provide guiding structural vocabulary for a team enacting a solution appropriate for a context laid out in the affinity diagram.



I. Label Making


Participant members Irene De La Torre, Jessie Richards, and Andrew Tang produced this affinity map under the facilitation of Patrick J. O’Donnel. The behavior explored was the “Pedestrians Crossing the Street,” ultimately fulfilling a larger study of waiting in urban contexts. Enumeration of actors, technologies, places, concepts, behaviors, interactions, emotions, and rationales plotted the range of data that comprises a pedestrian waiting and crossing the street. Post-it notes were given to each participant. As each concept was written down it was announced to the room, without concern of redundancy or judgment. The canvas filled up, and the generation of concepts slowed, after about seven minutes.

II. Label Grouping


During the silent phase, labels were moved frantically about. The most intriguing behavior was the use of space outside of the canvas for organization—like the wall the white board was attached to, and a nearby table top. Our canvas was full at the end of the label making, but white space between semi-grouped concepts seemed promote the cognitive function of organization while not overwhelming the visual search. A few participants found themselves picking up a label, trying to make it fit in, only to hand it off to another participant, in hopes they had a connection that sparked upon it being received. The result was nine groupings of concepts.

III. Chart Making


After reaching wa—a Japanese word for the harmony that arises from group consensus (Scupin 234)—the facilitator broke the silence with talks of categorization of the most obvious groupings first. To make titles for these groupings, the participants discussed each group individually. A natural reaction to a conceptually sound grouping was the instantaneous offer to suggest a title. If several iterations of a title could be suggested, more often the title was selected from them.

If the brainstorming of titles did not immediately yield results, the facilitator asked if there was an outlier in the group that is holding back a near-inclusive title. At this point, the creator of the outlier label gave a brief explanation as to why they included it initially. Usually this type of clarification guided the group to move the label to another grouping, or brainstorm an inclusive title. Once grouping titles were agreed upon, the post-it notes were circled by dry erase marker, to signal its completeness. The results for the pedestrian crossing diagram was category titles such as “actors in compromising safety,” “timing,” and “crosswalk awareness.”

IV. Analysis

The context charted by the affinity diagram shows a concern with human-recognized objects (actors and technology), actor concern for safety, and acts of crosswalk system awareness. Though the time spent waiting at a crosswalk is relatively low compared to the totality of an average pedestrian journey, a consciousness of system temporality was apparent. The participants also grouped system signals and distracting stimuli in a group called “crosswalk awareness” that suggests distraction and information-retention exist on a spectrum, and vary in attention from person to person or trip to trip. The amount of stimulation at a crosswalk can actively compete with each other. Additionally, many artifact and architectural labels were grouped, suggesting that actors are aware of the regularity/design of crosswalk intersections.

Pedestrians crossing the street is a context more nuanced than just an exchange between pedestrians and traffic that results in waiting. The improvisational behaviors of pedestrians making judgments about how to opportunistically jaywalk is a complex algorithm that is shaped by the “cognitive mind extension” described by Clark and Chalmers. The systems of trains, automobiles, cyclists and pedestrians all emit visual and sonic stimuli that, when subconsciously processed, is helpful—and life-saving—but, when attended to, is stress-inducing. Some might say these stimuli are even a symbolic epitome of hurriedness in urban life. The affinity diagramming brought about the considerations that constitute a context in which potential interventions and solutions can be embedded.


Beyer, Hugh. “User-centered agile methods.” Synthesis lectures on human-centered informatics 3.1 (2010): 1-71. Morgan & Claypool Publishers. Web. 17 Oct. 2016.

Clark, A. and Chalmers, D. “The extended mind.” Analysis 58, 1 (1998), 7–19.

Hoerl, Roger, and Ron D. Snee. Statistical thinking: improving business performance. Vol. 48. John Wiley & Sons, 2012. ProQuest. Web. 14 Oct. 2016.

Kawakita, Jiro. “The original KJ method.” Tokyo: Kawakita Research Institute (1991). Web. 17 Oct. 2016.

Scupin, Raymond. “The KJ Method: A Technique for Analyzing Data Derived from Japanese Ethnology.” Human organization 56.2 (1997): 233-7. ProQuest. Web. 14 Oct. 2016.




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s