Stumbling into a small peer support community
I arrived at Outlier 2021 eager to learn and explore. I was a few years into my transition from data scientist to visualization designer/developer and this was my first data viz conference.
Through the conference Slack, I connected with a handful of attendees who were looking to form a small peer support network. Most of us had started with R and were at different points on our journeys of learning d3 and other web-based visualization tools. Over the next few years, we met monthly and supported each other through career transitions, learning new technologies and approaches, and getting more involved in the DVS community.
For each of us, it was a period of rapid growth, and many of us found the group incredibly helpful in this period of transition. We plotted our growth in dataviz at the end of 2023.
A conference, together.
At the start of 2023, our group had seven members from seven different countries, and none of us had ever met in person before. Six of us were able to make the trip to Porto for Outlier 2023. There, we not only got to spend time together and deepen our friendships, but we also got to benefit from going through the conference as part of a cohort.
Having a bunch of familiar faces, each with our own handful of additional connections to other attendees, made the social graph of the conference feel so much less daunting. As someone who finds networking a challenge—and most conferences socially awkward—this was novel for me, and I had a great time!
I came out of it wondering if our experience was replicable. Could we form small groups with overlapping interests ahead of the conference, and through them make the conference feel more welcoming and build community?
Community Clusters at Outlier 2024
I pitched an idea to the Outlier 2024 team:
Step 1: Create a survey for attendees to express their interests ahead of the conference
Step 2: Use a clustering algorithm to group people with similar interests together
Step 3: Visualize the resulting Clusters to make people in each Cluster know what connects them
Step 4: Set up connection points for each Cluster—slack groups and time and space for in-person attendees to meet
The Outlier team liked the idea and “Community Clusters” were born.
For our survey, we asked about Topics of Interest, Industries/Issues. and Tools people wanted to base their clustering on.
Attendees showed more interest than I could have imagined. With just a few short messages over email and Slack, nearly two thirds of in-person attendees filled out the form and signed up for Clusters. While we had conceived of this as an in-person initiative, virtual attendees also were excited to cluster as well. In the end, we had more than 200 people in 25 Clusters.
For each Cluster, we looked at the top overlapping interests from the 40 survey questions, and how they deviated from conference attendees as a whole. Sophie Sparkes took this raw material and created finger-print inspired badges for each Cluster—where each line of the finger print represented one of the 40 topics in the survey; and I made more descriptive charts for each Cluster’s members to see their top overlapping interests.
A brief explainer of the clustering with digital fingerprints can be found here.
Results and learnings
While conference attendees loved the idea of Clusters, and some attendees found them to contribute significantly to their Outlier experience, the experiment overall fell short of our hopes.
We think this is for a combination of reasons, including:
1. Broad but shallow engagement. We ended up with five times as many Cluster participants as we initially planned for. But most people who signed up to join a Cluster didn’t ever post in their Slack group or come to an in-person meetup. Among those who filled out the Community Clusters evaluation survey, there was a strong correlation between how much people engaged with their Clusters and how likely they were to say Clusters strongly contributed to their Outlier experience; however, unfortunately many Clusters didn’t reach a critical mass of engagement to make interaction easy, especially online where only a handful of Clusters had more than a couple posts after initial introductions.
2. Logistic challenges. Many attendees didn’t join Slack ahead of the conference to meet and introduce themselves, the badges were hard to read in small format, and the venue layout made the meeting place hard to find.
We did not find conclusive evidence that the mechanics of the clustering algorithm were determinative of experience. Our original peer support group came together with a set of shared interests and overlapping strengths, which we found to be important in coming up with engaging topics for meetings and being able to grow together. As a result, we thought Clusters around certain types of interests, or with more shared interests, would be likely to lead to more connection. But this initiative didn’t give us enough data to meaningfully evaluate that. In general, we found people across the different Clusters thought their Cluster matched their interests relatively well; and we found that a handful of members in each Cluster—not necessarily those closest to the Cluster centroid, or those who had the most shared interest—were excited and engaged.
Next Steps
For the attendees who did find the Clusters to be a great experience it was fun to see how data, analytics, and visualization could work to build community. But we know there is so much more to be done.
As we look forward to Outlier 2025 and future gatherings, we’re excited to draw on lessons from this pilot. We know it’s probably not realistic to fully scale the experience of the peer support community I was in. But by forming clusters earlier, setting higher expectations of participation, and creating spaces for engagement ahead of the conference, we hope to narrow the gap between people’s desires for the program and their experiences.
Special thanks to…
- My fellow peer support group members – Including Carmen Torrecillas, Chris Woods, Maria Fedorova, Matthias Stahl, Patricia Tiffany Angkiriwang, and Stefan Pullen, whose data are represented above, as well as the other group members who have come and gone over the years.
- Duncan Geere for initially inspiring our group to form via his 2020 piece on community building. That piece reflected on his own experience of a small peer support community with Alli Torban, Will Chase, Gabrielle Merite; the group that eventually went on to form Elevate.
- The whole Outlier 2024 team, especially Zane Wolf who helped shape and encourage the Community Clusters initiative, Sophie Sparks for her wonderful badge designs, and Caroline Micoloi and the online volunteers she coordinated to support the Cluster Slack groups for virtual attendees.
