Data Quilt 2.0
Heer, Bostock, and Ogievetsky write, in “A Tour through the Visualization Zoo”, that “By making data more accessible and appealing, visual representations may also help engage more diverse audiences in exploration and analysis. The challenge is to create effective and engaging visualizations that are appropriate to the data.” (1) My final project for Information Visualization this semester is an analytic data quilt. This was the most appropriate choice to show a set of visualizations for this thematic data set, which is based on the portfolio of quilter and former slave Gracie Mitchell. Interviewed in 1938, as part of the WPA Federal Writers Project, her interviewer took notes while Gracie Mitchell showed her 30 different items, with 22 distinct quilt designs, and included a list of pattern names in her documentation. I used these 22 patterns, as well as time and location data for the interviewee, to filter data in the Quilt Index and identify the information that gives the audience for this visualization a context for understanding her design aesthetic. I decided to approach this infographic quilt top with three essential questions:
- How was she influenced by the quilt(er)s around her? She lived in Texas for ~40 years, and then Arkansas for 25 years, moved to Chicago for 8 years, and then returned to Arkansas.
- What is the geographical origin for each pattern? The provenance is known for many of the oldest surviving quilts, giving a sense of where patterns were first “published”.
- How popular were each of these designs during the final decades of slavery in the U.S., and during Gracie Mitchell’s lifetime? She might have been an innovator. Also, since she moved several times, Gracie Mitchell may have been a bridge in the network of quilters that existed and were sharing patterns in the late 19th and early 20th century.
The overall data quilt is designed to resemble a flag, and the colors used are hues of red, white, and blue, where red is consistently used to emphasize significant data, and the background white is a word cloud generated with Wordle from the RQP Digital Annotation exercise. I tried to pay attention to preattentive visual processing properties, as discussed by Christopher Healey, in “Perception in Visualization”, and use color for counting and estimation. I also limited myself, using only four web safe colors: 333366, 663333, 993333, and 666666, taking into consideration the red:green:blue 40:20:1 ratio discussed by Lindsay MacDonald in “Using Color Effectively in Computer Graphics”, and deciding to omit green so red-green colorblind individuals will not have difficulty. Reading from left to right, top to bottom, network visualization is encountered first. The layout I selected for this ego network is a circle; it is appropriate for the first question (above), and the audience will associate it with variations of the American flag during and directly after the time of Reconstruction:
I placed Gracie’s 3 geographic locations in the center, surrounded by a circle of nodes representing the 22 quilt designs she executed, and an outer circle of nodes representing the designs (of surviving quilts) that were made in the places she lived, while she was there, by other quilters. Each edge represents the quilter who made the item, and dashed lines represent Gracie Mitchell. I used Sawtooth Star icons for the nodes, which might be considered “chart junk” by Scott Bateman, but they are accurate interpretations of stars on quilt tops (which usually have 8 points), and their memorability contributes to the impression of an American flag quilt. The star nodes are sized according to in-degree, and for the edges between nodes I used grayscale to code the date the quilt was made, for example black edges represent quilts made during the era of slavery.
Geographic data for Gracie Mitchell’s 22 designs, where color codes the date the pattern was “published”, is displayed with small multiples set in rows to the right of the network, giving the impression of stripes. I used square markers on a grayscale U.S. map to show the first 1-10 instances of a quilt design, based on the surviving quilts in the Quilt Index. The size of the square codes frequency, red indicates an older pattern, and blue indicates patterns that were “published” later, during Gracie’s lifetime.
Finally, a heat map runs across the bottom of the data quilt top, adding to the flag impression with more horizontal stripes. The heat map shows the frequency with which 21 of the 22 patterns were made; the log cabin pattern occurs the most frequently (making up almost half of the items retrieved for the data set), so it is highlighted with its own chart, separated from the heat map because it is an outlier, and heavily skews the visualization when included. In the heat map time is shown in 5 year buckets, a vibrant red codes high frequency, and dark blue codes low frequency. In the chart for the log cabin, like the small multiple maps, red codes an item as older, blue as newer, and size codes frequency.
Overall this data quilt represents using data to place historical figure Gracie Mitchell in context. The visualizations show that she was experimental, and executed several quilt designs that were established patterns, but not frequently made by her contemporaries. She also created a few quilt tops that demonstrate her familiarity with traditional patterns that existed during the era of slavery, and which continued to be popular throughout her lifetime. Michael Friendly notes, in “A Brief History of Data Visualization”, that the beginning of modern graphics took place 1800-1850, a period when American quilters were still developing many of the block patterns now considered traditional and they passed directly from one quilter to another. Likewise, the “Golden Age of Statistical Graphics” (1850-1900) mirrors a period of innovation in quilting; traditional quilt block patterns began to get frequently published in print, in ladies magazines and national newsletters, passing from one quilter to many. The printing press and the industrialized textile mill both contributed to a “Golden Age” for American quilting, followed by a dark age (caused by two World Wars, and women joining the workforce), a rebirth (the American Bicentennial quilt revival), and the modern day quilting practice was revolutionized by the invention of the rotary cutter in 1979, around the same time that data visualization was positively impacted by access to personal computers.
In “The Value of Information Visualization”, Fekete, Wijk, Stasko, and North identify situations where browsing data works for an exploratory task, such as my Runaway Quilt Project:
- There is an underlying structure: Brackman numbers are a classification system for quilt patterns, which is used by quilters, museums, archives, and scholars. Established with the publishing of her Encyclopedia of Pieced Quilt Patterns in 1993, this organization scheme offers authoritative vocabulary and a hierarchy defined with numbers. I was able to use the Brackman Encyclopedia to identify the authoritative names of Gracie Mitchell’s chosen quilt designs. I cross-referenced the names and patterns by searching the Quilt Index using Brackman numbers, authority terms, and the quilter’s natural language.
- Users are unfamiliar with the content: An audience of quilters is may be unfamiliar with information visualization, and an audience of informationists may be unfamiliar with the history of quilting. This final project speaks to a diverse range of scholars and laywomen, and it is an interesting way of presenting both fields to a general audience.
- The information is easier to recognize than describe: Quilt patterns have descriptive names, but Gestalt theory, as described by Fekete et al, maintains that there is insight to be gained from proximity, so I set each authoritative block pattern next to a map showing the time and location of its origin. In the future, I would include more data points (to show how the patterns spread over time) and have a time-lapse animation on the research blog. If I was able to redesign the quilt, I might rearrange the blocks from oldest published forward. Right now the pattern names in the visualizations are ordered alphabetically or by provenance (see Gracie Mitchell’s transcript).
After designing the visualizations for the quilt top on a computer, digital patchwork, a friend (and Pratt Industrial Design alum) created original fabric designs inspired by the terms “woman”, “make”, “cotton”, and “warm” – because a quilt is something a woman makes out of cotton to keep you warm –for large scale renditions of each block design which make up a “cheater cloth” for the reverse side of the data quilt. Another friend helped me prepare the final infographic file in Adobe InDesign for printing on to fabric at Pratt’s Digital Output Center on the Brooklyn campus. I quilted the final tactile object at home, by machine, and then submitted this quilt, and “Maker Unknown” (my final project for 697 Digital Humanities last spring), to the International Quilt Festival for an exhibit titled “In the American Tradition”.
