New month, new update. Today I am writing about the process of designing my visualizations/quilt blocks. If you are new here you can read my previous posts: an introduction to the project and an update about my data.
Designing an Embodied Visualization
When I first conceived of this project I had a pretty simple design in mind. I was planning on taking a tree map of my data, generated in Tableau, and copying those dimensions in fabric. I was then thinking about ways of layering more data onto this simpler visualization, perhaps embroidering more common objects in the dataset and using hand quilted lines to show the number of words used for each color. But ultimately I moved away from this plan, I wanted to think more about how I could translate my data into a quilt, rather than just copying a computer generated visualization. How could I use the traditional elements of a quilt, typically equally sized square quilt blocks, to represent my data?
This was happening at the same time I was working with my data, and as I described in my data post, I decided it was more interesting to think about what the colors in my data were describing rather than just how many colors there were in total. By categorizing those objects, I found a more compelling visual narrative. So rather than making one large visualization, I made a quilt block for each category, thus translating my data into a more traditional quilt form AND turning my quilt into a data dashboard.
While quilt blocks are traditional uniform in size and similar in design, due to the differing number of instances of each category that was not going to be possible since I didn’t want to make it based on percentages. Instead the size of the block shows how common, or uncommon, that category is in the text. This also meant each block could not have the same pattern, so instead I designed each block individually.
Before I designed the blocks I needed to determine what my units were. I decided my base unit would be one instance equaled one 2 inch square. However if I only used squares the patterns weren’t going to be that interesting, so I allowed myself to also use half square triangles (HST) where 2 HST equaled one instance as well. HST are a classic element of quilt designs and this gave me more artistic freedom.
The only remaining issue was that each category wasn’t necessarily a square number, so I allowed myself another element. I added a “null” color, in this case light gray, that was also going to be used for the borders between the blocks. This allowed me to make rectilinear blocks. I did add a few restraints, I tried to limit how many null squares I could add and I had to keep the null squares on the edges of the block. This way once the borders were added they would hopefully fade into the background and look less like it is part of the visualization.
I finished designing the quilt top while I was working with my data, but I look forward to continuing to think about the design, because that process is not over. I want to use every available element of the quilt to represent data. When thinking about a quilt there are a few key elements: the quilt top, the quilt backing, the stitches that attach the two sides (and battling) together, the binding on the edge of the quilt. I have a few ideas, but I am trying to not rush this process and take time to make the final decision. I wound up using more of the light gray fabric than I anticipated, so I’ve ordered more, and I am taking a break from this part of the project as I wait for it to arrive.
Sewing As Data Work
Nicole Cote, in her incredible repository of data visualization resources VisDepot, makes the case that data sketching should be an integral part of the visualization process. Before you generate a visualization it is a great practice to brainstorm on paper what you want your visualizations to be. Otherwise I think it is too easy to get stuck generating the same types of visualizations over and over again, because they are the ones you know. The sketching stage of this project was so invaluable for the final product. I was able to do so much of my thinking and analysis by spending dedicated time sketching and experimenting with designs. This isn’t a step I always have taken in the past, but it is one I am realizing the importance of more and more. This project just highlighted that lesson for me.
I recorded a video of the process of creating one of the quilt blocks to show during my DH Unbound presentation. I thought it would be a more interesting way to show people the process than showing photos or describing the work. One of the outcomes of even just editing the video was I realized that at least 80% of sewing isn’t actually sewing. The actual act of sewing might only be like 5% of sewing. So much of the process is measuring, cutting, trimming, and ironing the fabric before you can even start sewing. And once you start sewing you still have to be ironing and other maintenance tasks regularly. I feel like this is a strong parallel for what it’s like to visualize data. Most of the work is gathering, tidying, and fixing data. The act of visualizing the data is only a small part of the process. In this way a data sewing project, or other data craft, could be a great tool to teach the labor aspects of data work. I believe embodied data has a lot of potential for data pedagogy, and this labor parallel is just one way it can help us teach about data.
Updates and Next Steps
- I am going to update my data and documentation in my repository to reflect new terminology. At some point I decided “people” was a more accurate descriptor than “bodies” as bodies felt like it was implying corpses, when in fact most of the people were still alive and described things like hair. I made this change after I uploaded my materials, so I need to update that.
- As I’ve mentioned previously, I am presenting this project in May at DH Unbound. As I am taking a pause from the sewing/designing of the quilt I want to use the next few weeks to wrap my head around the academic theories and projects that are inspiring my process. I think my next post will actually be the script of my talk rather than a normal update, and I want to use the time to really think about the narrative of my work.
Cote, Nicole. (2021). VisDepot: An Introductory Resource for Data Visualization, v1.0.1. https://doi.org/10.5281/zenodo.5234286