The use of affinity diagrams in your interpretation sessions aims to help you to develop insights. By organizing and categorizing the evidence you collected, you can uncover the needs, frustrations and values of your informants, guided by actual observations rather than by your prior assumptions. Here are a few tips for using this methodology effectively:

1. Before the affinity diagram — the raw data: Your insights will always be limited by the data you collected, so there is really no substitution to having good raw data. Gather as many observations and you can, and clearly describe each observation on a note. Observations should be concise (e.g., “tried to get into a lecture but got kicked out”, but not too concise (e.g., “lecture”). For each observation, also record basic information about the informant who generated it (e.g.  “tried to get into a lecture but got kicked out” [20 y/o student from Brazil]). If you have certain demographic groups you can also consider using different colors of post-its/sharpies to indicate different groups.
2. Categorization — what is a good category? Reflect on your observations, and try to group them by the underlying needs, frustrations and values that they uncover, rather than by activity or general topic. For example, “want to engage in Harvard student life” is a good category because it describes a desire of your informants; “photos” is not a useful category because it doesn’t tell us much about the what our informants care about. Different observations related to tourists taking photos might reveal very different underlying values/needs/frustrations: the observation “tourist took a picture of a lecture” might suggest that the tourist was curious about student life at Harvard and/or that it’s important for them to document their experiences, while the observation “took a picture of the sign by the John Harvard sculpture and read the information on his phone” might suggest that the person was interested in learning about the history of Harvard, and perhaps also that the location of the sign was uncomfortable for reading or was crowded. (Remember, you’ll know what’s the right interpretation only by asking “why?” a lot…). If you come up with good categories, you essentially have your insights.
3. Categorization — how to find categories: Coming up with good categories is non-trivial and typically requires an iterative process. When reading an observation you might want to recall more about the context (which was maybe not entirely documented on the post-it), you can try to look back in your notes and expand the observation. Generally, a single observation might relate to more than one value/need/frustration. Therefore, it is a good idea to try out multiple categorizations (for example, observations #1, #3 and #5 might reveal one need, but observation #1 together with #8 and #10 might reveal another frustration). Of course, every time you think you have a good category, document it before you take it apart… If you get stuck, you can try to pick an observation at random and see if anything else relates to it in some way (or pick a group at random and see if a subset fits in some interesting way). Also take breaks, incubation can be helpful and starting fresh can help you get “unstuck”.