Hi, all.
Happy optional module week to those who celebrate.
After going back and forth on which module to choose (and a short-lived attempt to teach myself network analysis methods), I settled on the data cleaning and analysis module. Before we experiment with more data visualization methods in this class, I wanted to learn more about these magical .csv files we have been downloading and using these past couple of weeks.
Going into the data module, I didn’t realize I would find myself immersed in a scrutinizing debate spanning over ten years over what data in the humanities should be called. Here’s my summary of the drama:
It’s 2011. Johanna Drucker disrupts the DH world with “Humanities Approaches to Graphical Display,” an article in which she calls for a fundamental change in how humanists understand the data they use and how it affects the visualizations they create. Drucker makes a convincing case, arguing that humanists should never treat the scientific idea of “data” as objective, empirical fact, and that we must consider the constructedness and subjectivity of records and information across geographic, mathematical, and scientific information, not just first-person primary sources. This all seems reasonable, right?
Then there’s the weird part. Drucker’s call to action is to rename and 〰reconceptualize〰 all organized information in the digital humanities as something fundamentally different than traditional data: “capta.”1
EDITOR’S NOTE: At this point in my reading, I’m thinking to myself: yeah, okay, Dr. Drucker. I’m with you all the way up until this one. Since Drucker herself argues that all data should be understood as “capta,” or human-modified information, she can’t push humanists to use a separate term without totally isolating the field in the process.
ten… years… later… (or ten… seconds… as I wait… for the next article to load…)
It’s 2021, the tenth anniversary of Drucker’s article. To celebrate, Matthew Lavin publishes his response to Drucker entitled “Why Digital Humanists Should Emphasize Situated Data over Capta.” Lavin is obviously also not a fan of Drucker’s “capta” proposal. Although he agrees with Drucker’s sentiments on humanistic data use, he argues that we actually shouldn’t make the full switch from “data” to “capta” for a variety of reasons, many of them practical, including the difficulties it would create when working across disciplinary lines and with the public. In sum, Lavin thinks no one would know what “capta” means, and the time/space/effort it takes to explain a personal embrace of alternative terms to those with widely accepted conventions would be better expended in further exploring their methodology and interpretations of the information.2
Instead of capta, Lavin suggests humanists use qualifying words to signal the subjective nature of their source information, such as “situated data.” But even then they should never, in any case, shy away from using the word data, as it would only perpetuate the false distinction between “subjective” humanistic information and “objective” scientific information. Basically, his response was:
“Stop trying to make ‘capta’ happen. It’s not going to happen.”
So yes, ultimately the debate ends where it was before Drucker stepped in: with humanists continuing to use the term data for their organized information. Who would have thought?
With Great Data Comes Great Responsibility…?
More on the importance of responsible data use from Clodomir Santana et al.’s article, including thorough documentation of all data processing steps to better ensure scientific replicability. This includes the state of the raw data, all ways that data was edited, and the exact conditions under which it was transformed or visualized. Santana et al. also discuss using a computer program to read and extract metadata from large sets of source documents.3 This idea seemed really useful to me, especially to parse through large archival collections, but I genuinely would have no idea where to start with programming my own code. Maybe one day…?
Tableau shared helpful tips for data cleaning, a process essential to combining information from various sources into one cohesive dataset. Their suggestions include removing and reducing duplicate entries, fixing structural and mislabeling errors in rows/columns/categories, and identifying outliers in your data. They say to “filter” these outliers, but in a humanist setting I suggest you “consider” them. Why may they have deviated from the norm?4
Playing with OpenRefine was fun once I figured out I needed to manually save spreadsheets as .csv files instead of uploading the automatic .numbers format Mac downloads. When filtering the sample dataset, I found 160 priests and 35 teachers, 16 midwives, and just one “doctor.” I thought this was strange, then realized firsthand that keyword searching from a modern perspective can skew our understanding of past societies; I eventually filtered for “physician,” and found 424.
For my part, I uploaded a spreadsheet with existing metadata from the 100 Years of Black Falls Church collection that we were recently sent to edit. This 419-row index was the closest thing I had to “vast” amounts of historical data, so I probably didn’t get the full experience this software can offer. I found that I could filter item descriptions to find references to people and topics, such as all references to Louisa Henderson and her store. I also facet searched by creator to determine where the items in the archive originated. This search highlighted the key sources: the Henderson and Tinner Family Collections, and illuminated Dr. Manuel Scott’s extensive work on the project ten years ago. Facet searching the genre category revealed 273 uncategorized entries (over half!) that don’t fit under the existing “sound” or “still image” genres, even though you wouldn’t know it from looking at the top rows of data. My group hopes to fill in some of that missing metadata this semester!
Until next week,
Alexa
- Johanna Drucker, “Humanities Approaches to Graphical Display,” Digital Humanities Quarterly, 5, no. 1 (2011), https://dhq.digitalhumanities.org/vol/5/1/000091/000091.html. ↩︎
- Matthew Lavin, “Why Digital Humanists Should Emphasize Situated Data over Capta,” Digital Humanities Quarterly, 15, no. 2 (2021), https://dhq.digitalhumanities.org/vol/15/2/000556/000556.html. ↩︎
- Clodomir Santana et al., “Unveiling the Critical Nexus of Data Preprocessing and Transparent Documentation for Result Quality and Reproducibility in Digital History,” Digital Humanities Quarterly 19, no. 2 (2025), https://dhq.digitalhumanities.org/vol/19/2/000788/000788.html. ↩︎
- Tableau, “Guide to Data Cleaning: Definition, Benefits, Components, and How to Clean Your Data,” https://www.tableau.com/learn/articles/what-is-data-cleaning.
↩︎
Be First to Comment